Upload fine-tuned model
Browse files- config.json +40 -0
- generation_config.json +12 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scaler.pt +3 -0
- scheduler.pt +3 -0
- trainer_state.json +2493 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"LlamaForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_bias": false,
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"bos_token_id": 128000,
|
| 8 |
+
"eos_token_id": [
|
| 9 |
+
128001,
|
| 10 |
+
128008,
|
| 11 |
+
128009
|
| 12 |
+
],
|
| 13 |
+
"head_dim": 64,
|
| 14 |
+
"hidden_act": "silu",
|
| 15 |
+
"hidden_size": 2048,
|
| 16 |
+
"initializer_range": 0.02,
|
| 17 |
+
"intermediate_size": 8192,
|
| 18 |
+
"max_position_embeddings": 131072,
|
| 19 |
+
"mlp_bias": false,
|
| 20 |
+
"model_type": "llama",
|
| 21 |
+
"num_attention_heads": 32,
|
| 22 |
+
"num_hidden_layers": 16,
|
| 23 |
+
"num_key_value_heads": 8,
|
| 24 |
+
"pad_token_id": 128256,
|
| 25 |
+
"pretraining_tp": 1,
|
| 26 |
+
"rms_norm_eps": 1e-05,
|
| 27 |
+
"rope_scaling": {
|
| 28 |
+
"factor": 32.0,
|
| 29 |
+
"high_freq_factor": 4.0,
|
| 30 |
+
"low_freq_factor": 1.0,
|
| 31 |
+
"original_max_position_embeddings": 8192,
|
| 32 |
+
"rope_type": "llama3"
|
| 33 |
+
},
|
| 34 |
+
"rope_theta": 500000.0,
|
| 35 |
+
"tie_word_embeddings": true,
|
| 36 |
+
"torch_dtype": "float32",
|
| 37 |
+
"transformers_version": "4.51.3",
|
| 38 |
+
"use_cache": true,
|
| 39 |
+
"vocab_size": 128257
|
| 40 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 128000,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
128001,
|
| 6 |
+
128008,
|
| 7 |
+
128009
|
| 8 |
+
],
|
| 9 |
+
"temperature": 0.6,
|
| 10 |
+
"top_p": 0.9,
|
| 11 |
+
"transformers_version": "4.51.3"
|
| 12 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9fdc270d0794881d3e4a0ed7d3b6105765b1e7e3e16922d98612e3bfb969c5ac
|
| 3 |
+
size 4943282520
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8f726b03a2853fc2c8f649e1c3a0f6c78a946a1b6eaf80acfcdc37afb30f023d
|
| 3 |
+
size 2510811426
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3c43c644ac2638b1bebf307075b7a121ea070bbc2a96c435e274fa0feacaf39
|
| 3 |
+
size 14168
|
scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:31488f9db4720708cb6ca4d01975619b5c575bec7c8255374e3c64a0f7464a1b
|
| 3 |
+
size 992
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:06a9c405d807dc88acd9198f1ae88c285facf0c59873477061f372d72da37143
|
| 3 |
+
size 1056
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2493 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 3.9990416866315286,
|
| 6 |
+
"eval_steps": 200,
|
| 7 |
+
"global_step": 3336,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.011978917105893628,
|
| 14 |
+
"grad_norm": 476.6510925292969,
|
| 15 |
+
"learning_rate": 8e-07,
|
| 16 |
+
"loss": 11.6475,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.023957834211787255,
|
| 21 |
+
"grad_norm": 74.57940673828125,
|
| 22 |
+
"learning_rate": 1.9999928625229307e-06,
|
| 23 |
+
"loss": 2.3869,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.035936751317680884,
|
| 28 |
+
"grad_norm": 125.54178619384766,
|
| 29 |
+
"learning_rate": 1.999912567076008e-06,
|
| 30 |
+
"loss": 7.1899,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.04791566842357451,
|
| 35 |
+
"grad_norm": 14.804201126098633,
|
| 36 |
+
"learning_rate": 1.999743061523497e-06,
|
| 37 |
+
"loss": 5.0722,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.059894585529468136,
|
| 42 |
+
"grad_norm": 9.310958862304688,
|
| 43 |
+
"learning_rate": 1.999484360988329e-06,
|
| 44 |
+
"loss": 2.9189,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.07187350263536177,
|
| 49 |
+
"grad_norm": 306.2783508300781,
|
| 50 |
+
"learning_rate": 1.999136488551224e-06,
|
| 51 |
+
"loss": 2.8403,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.08385241974125539,
|
| 56 |
+
"grad_norm": 134.76495361328125,
|
| 57 |
+
"learning_rate": 1.9986994752486316e-06,
|
| 58 |
+
"loss": 4.2047,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.09583133684714902,
|
| 63 |
+
"grad_norm": 78.88103485107422,
|
| 64 |
+
"learning_rate": 1.998173360069964e-06,
|
| 65 |
+
"loss": 5.1269,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.10781025395304264,
|
| 70 |
+
"grad_norm": 60.88028335571289,
|
| 71 |
+
"learning_rate": 1.997558189954117e-06,
|
| 72 |
+
"loss": 4.8787,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.11978917105893627,
|
| 77 |
+
"grad_norm": 48.85028076171875,
|
| 78 |
+
"learning_rate": 1.9968540197852784e-06,
|
| 79 |
+
"loss": 2.6971,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.1317680881648299,
|
| 84 |
+
"grad_norm": 82.52726745605469,
|
| 85 |
+
"learning_rate": 1.9960609123880376e-06,
|
| 86 |
+
"loss": 6.6349,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.14374700527072354,
|
| 91 |
+
"grad_norm": 18.934968948364258,
|
| 92 |
+
"learning_rate": 1.9951789385217753e-06,
|
| 93 |
+
"loss": 3.6926,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.15572592237661714,
|
| 98 |
+
"grad_norm": 95.94075012207031,
|
| 99 |
+
"learning_rate": 1.9942081768743535e-06,
|
| 100 |
+
"loss": 5.221,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.16770483948251078,
|
| 105 |
+
"grad_norm": 69.58992767333984,
|
| 106 |
+
"learning_rate": 1.9931487140550935e-06,
|
| 107 |
+
"loss": 5.8621,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.1796837565884044,
|
| 112 |
+
"grad_norm": 18.388065338134766,
|
| 113 |
+
"learning_rate": 1.9920006445870497e-06,
|
| 114 |
+
"loss": 5.2103,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.19166267369429804,
|
| 119 |
+
"grad_norm": 117.37078857421875,
|
| 120 |
+
"learning_rate": 1.9907640708985766e-06,
|
| 121 |
+
"loss": 5.8106,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.20364159080019167,
|
| 126 |
+
"grad_norm": 118.88228607177734,
|
| 127 |
+
"learning_rate": 1.9894391033141887e-06,
|
| 128 |
+
"loss": 4.0891,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.21562050790608528,
|
| 133 |
+
"grad_norm": 17.98489761352539,
|
| 134 |
+
"learning_rate": 1.9880258600447204e-06,
|
| 135 |
+
"loss": 5.7061,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.2275994250119789,
|
| 140 |
+
"grad_norm": 170.75030517578125,
|
| 141 |
+
"learning_rate": 1.986524467176777e-06,
|
| 142 |
+
"loss": 4.2787,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.23957834211787254,
|
| 147 |
+
"grad_norm": 74.85968017578125,
|
| 148 |
+
"learning_rate": 1.9849350586614863e-06,
|
| 149 |
+
"loss": 7.8201,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.23957834211787254,
|
| 154 |
+
"eval_loss": 1.2861307859420776,
|
| 155 |
+
"eval_runtime": 238.5238,
|
| 156 |
+
"eval_samples_per_second": 6.247,
|
| 157 |
+
"eval_steps_per_second": 3.123,
|
| 158 |
+
"step": 200
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.25155725922376615,
|
| 162 |
+
"grad_norm": 254.58087158203125,
|
| 163 |
+
"learning_rate": 1.983257776302548e-06,
|
| 164 |
+
"loss": 5.8449,
|
| 165 |
+
"step": 210
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.2635361763296598,
|
| 169 |
+
"grad_norm": 12.087563514709473,
|
| 170 |
+
"learning_rate": 1.9814927697435826e-06,
|
| 171 |
+
"loss": 5.7451,
|
| 172 |
+
"step": 220
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 0.2755150934355534,
|
| 176 |
+
"grad_norm": 11.869245529174805,
|
| 177 |
+
"learning_rate": 1.9796401964547794e-06,
|
| 178 |
+
"loss": 6.4206,
|
| 179 |
+
"step": 230
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.2874940105414471,
|
| 183 |
+
"grad_norm": 18.04955291748047,
|
| 184 |
+
"learning_rate": 1.977700221718848e-06,
|
| 185 |
+
"loss": 3.3466,
|
| 186 |
+
"step": 240
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.2994729276473407,
|
| 190 |
+
"grad_norm": 40.88437271118164,
|
| 191 |
+
"learning_rate": 1.975673018616273e-06,
|
| 192 |
+
"loss": 4.5986,
|
| 193 |
+
"step": 250
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.3114518447532343,
|
| 197 |
+
"grad_norm": 17.213830947875977,
|
| 198 |
+
"learning_rate": 1.97355876800987e-06,
|
| 199 |
+
"loss": 3.6809,
|
| 200 |
+
"step": 260
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.32343076185912795,
|
| 204 |
+
"grad_norm": 31.727243423461914,
|
| 205 |
+
"learning_rate": 1.9713576585286513e-06,
|
| 206 |
+
"loss": 4.692,
|
| 207 |
+
"step": 270
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.33540967896502155,
|
| 211 |
+
"grad_norm": 95.36638641357422,
|
| 212 |
+
"learning_rate": 1.9690698865509964e-06,
|
| 213 |
+
"loss": 6.1814,
|
| 214 |
+
"step": 280
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.3473885960709152,
|
| 218 |
+
"grad_norm": 25.670534133911133,
|
| 219 |
+
"learning_rate": 1.966695656187131e-06,
|
| 220 |
+
"loss": 2.8556,
|
| 221 |
+
"step": 290
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.3593675131768088,
|
| 225 |
+
"grad_norm": 137.57431030273438,
|
| 226 |
+
"learning_rate": 1.9642351792609162e-06,
|
| 227 |
+
"loss": 3.3607,
|
| 228 |
+
"step": 300
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 0.3713464302827024,
|
| 232 |
+
"grad_norm": 46.084354400634766,
|
| 233 |
+
"learning_rate": 1.9616886752909523e-06,
|
| 234 |
+
"loss": 6.1352,
|
| 235 |
+
"step": 310
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.3833253473885961,
|
| 239 |
+
"grad_norm": 110.49552154541016,
|
| 240 |
+
"learning_rate": 1.9590563714709916e-06,
|
| 241 |
+
"loss": 5.8323,
|
| 242 |
+
"step": 320
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.3953042644944897,
|
| 246 |
+
"grad_norm": 77.74449157714844,
|
| 247 |
+
"learning_rate": 1.9563385026496687e-06,
|
| 248 |
+
"loss": 5.7407,
|
| 249 |
+
"step": 330
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.40728318160038335,
|
| 253 |
+
"grad_norm": 26.705305099487305,
|
| 254 |
+
"learning_rate": 1.9535353113095493e-06,
|
| 255 |
+
"loss": 5.3508,
|
| 256 |
+
"step": 340
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 0.41926209870627695,
|
| 260 |
+
"grad_norm": 146.0390167236328,
|
| 261 |
+
"learning_rate": 1.9506470475454957e-06,
|
| 262 |
+
"loss": 2.9407,
|
| 263 |
+
"step": 350
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.43124101581217056,
|
| 267 |
+
"grad_norm": 25.178991317749023,
|
| 268 |
+
"learning_rate": 1.947673969042353e-06,
|
| 269 |
+
"loss": 3.0089,
|
| 270 |
+
"step": 360
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 0.4432199329180642,
|
| 274 |
+
"grad_norm": 8.458102226257324,
|
| 275 |
+
"learning_rate": 1.9446163410519603e-06,
|
| 276 |
+
"loss": 2.885,
|
| 277 |
+
"step": 370
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.4551988500239578,
|
| 281 |
+
"grad_norm": 94.5208740234375,
|
| 282 |
+
"learning_rate": 1.9414744363694842e-06,
|
| 283 |
+
"loss": 3.8878,
|
| 284 |
+
"step": 380
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.4671777671298515,
|
| 288 |
+
"grad_norm": 138.86561584472656,
|
| 289 |
+
"learning_rate": 1.938248535309083e-06,
|
| 290 |
+
"loss": 5.5948,
|
| 291 |
+
"step": 390
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.4791566842357451,
|
| 295 |
+
"grad_norm": 10.55215072631836,
|
| 296 |
+
"learning_rate": 1.9349389256788943e-06,
|
| 297 |
+
"loss": 2.9242,
|
| 298 |
+
"step": 400
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.4791566842357451,
|
| 302 |
+
"eval_loss": 1.1180063486099243,
|
| 303 |
+
"eval_runtime": 237.218,
|
| 304 |
+
"eval_samples_per_second": 6.281,
|
| 305 |
+
"eval_steps_per_second": 3.141,
|
| 306 |
+
"step": 400
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 0.4911356013416387,
|
| 310 |
+
"grad_norm": 11.329914093017578,
|
| 311 |
+
"learning_rate": 1.931545902755359e-06,
|
| 312 |
+
"loss": 5.7209,
|
| 313 |
+
"step": 410
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 0.5031145184475323,
|
| 317 |
+
"grad_norm": 121.2057113647461,
|
| 318 |
+
"learning_rate": 1.928069769256879e-06,
|
| 319 |
+
"loss": 4.2294,
|
| 320 |
+
"step": 420
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.515093435553426,
|
| 324 |
+
"grad_norm": 70.16046905517578,
|
| 325 |
+
"learning_rate": 1.9245108353168055e-06,
|
| 326 |
+
"loss": 5.1172,
|
| 327 |
+
"step": 430
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.5270723526593196,
|
| 331 |
+
"grad_norm": 55.029964447021484,
|
| 332 |
+
"learning_rate": 1.9208694184557735e-06,
|
| 333 |
+
"loss": 3.8455,
|
| 334 |
+
"step": 440
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.5390512697652132,
|
| 338 |
+
"grad_norm": 16.75533103942871,
|
| 339 |
+
"learning_rate": 1.9171458435533706e-06,
|
| 340 |
+
"loss": 2.1762,
|
| 341 |
+
"step": 450
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.5510301868711068,
|
| 345 |
+
"grad_norm": 37.192169189453125,
|
| 346 |
+
"learning_rate": 1.913340442819153e-06,
|
| 347 |
+
"loss": 4.6994,
|
| 348 |
+
"step": 460
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.5630091039770004,
|
| 352 |
+
"grad_norm": 190.2852020263672,
|
| 353 |
+
"learning_rate": 1.9094535557630067e-06,
|
| 354 |
+
"loss": 8.188,
|
| 355 |
+
"step": 470
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.5749880210828942,
|
| 359 |
+
"grad_norm": 14.840240478515625,
|
| 360 |
+
"learning_rate": 1.905485529164856e-06,
|
| 361 |
+
"loss": 2.4346,
|
| 362 |
+
"step": 480
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.5869669381887878,
|
| 366 |
+
"grad_norm": 17.85882568359375,
|
| 367 |
+
"learning_rate": 1.9014367170437255e-06,
|
| 368 |
+
"loss": 5.1088,
|
| 369 |
+
"step": 490
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.5989458552946814,
|
| 373 |
+
"grad_norm": 98.2167739868164,
|
| 374 |
+
"learning_rate": 1.8973074806261558e-06,
|
| 375 |
+
"loss": 4.4192,
|
| 376 |
+
"step": 500
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 0.610924772400575,
|
| 380 |
+
"grad_norm": 72.538330078125,
|
| 381 |
+
"learning_rate": 1.8930981883139734e-06,
|
| 382 |
+
"loss": 4.2753,
|
| 383 |
+
"step": 510
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 0.6229036895064686,
|
| 387 |
+
"grad_norm": 121.5967788696289,
|
| 388 |
+
"learning_rate": 1.8888092156514252e-06,
|
| 389 |
+
"loss": 5.0462,
|
| 390 |
+
"step": 520
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 0.6348826066123623,
|
| 394 |
+
"grad_norm": 161.8177947998047,
|
| 395 |
+
"learning_rate": 1.8844409452916719e-06,
|
| 396 |
+
"loss": 3.2489,
|
| 397 |
+
"step": 530
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 0.6468615237182559,
|
| 401 |
+
"grad_norm": 149.68197631835938,
|
| 402 |
+
"learning_rate": 1.8799937669626481e-06,
|
| 403 |
+
"loss": 4.8399,
|
| 404 |
+
"step": 540
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 0.6588404408241495,
|
| 408 |
+
"grad_norm": 87.29440307617188,
|
| 409 |
+
"learning_rate": 1.8754680774322934e-06,
|
| 410 |
+
"loss": 5.3579,
|
| 411 |
+
"step": 550
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.6708193579300431,
|
| 415 |
+
"grad_norm": 70.05744171142578,
|
| 416 |
+
"learning_rate": 1.8708642804731513e-06,
|
| 417 |
+
"loss": 1.967,
|
| 418 |
+
"step": 560
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.6827982750359367,
|
| 422 |
+
"grad_norm": 49.84896469116211,
|
| 423 |
+
"learning_rate": 1.866182786826347e-06,
|
| 424 |
+
"loss": 4.1978,
|
| 425 |
+
"step": 570
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.6947771921418304,
|
| 429 |
+
"grad_norm": 29.61354637145996,
|
| 430 |
+
"learning_rate": 1.861424014164941e-06,
|
| 431 |
+
"loss": 4.025,
|
| 432 |
+
"step": 580
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 0.706756109247724,
|
| 436 |
+
"grad_norm": 99.13072204589844,
|
| 437 |
+
"learning_rate": 1.8565883870566666e-06,
|
| 438 |
+
"loss": 4.1162,
|
| 439 |
+
"step": 590
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 0.7187350263536176,
|
| 443 |
+
"grad_norm": 130.23606872558594,
|
| 444 |
+
"learning_rate": 1.8516763369260492e-06,
|
| 445 |
+
"loss": 3.0065,
|
| 446 |
+
"step": 600
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.7187350263536176,
|
| 450 |
+
"eval_loss": 1.104053258895874,
|
| 451 |
+
"eval_runtime": 238.4518,
|
| 452 |
+
"eval_samples_per_second": 6.249,
|
| 453 |
+
"eval_steps_per_second": 3.124,
|
| 454 |
+
"step": 600
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 0.7307139434595112,
|
| 458 |
+
"grad_norm": 50.47702407836914,
|
| 459 |
+
"learning_rate": 1.8466883020159161e-06,
|
| 460 |
+
"loss": 4.3503,
|
| 461 |
+
"step": 610
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.7426928605654048,
|
| 465 |
+
"grad_norm": 17.21928596496582,
|
| 466 |
+
"learning_rate": 1.8416247273482988e-06,
|
| 467 |
+
"loss": 4.4346,
|
| 468 |
+
"step": 620
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 0.7546717776712986,
|
| 472 |
+
"grad_norm": 49.47705841064453,
|
| 473 |
+
"learning_rate": 1.8364860646847262e-06,
|
| 474 |
+
"loss": 3.9906,
|
| 475 |
+
"step": 630
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.7666506947771922,
|
| 479 |
+
"grad_norm": 14.143331527709961,
|
| 480 |
+
"learning_rate": 1.831272772485922e-06,
|
| 481 |
+
"loss": 3.3026,
|
| 482 |
+
"step": 640
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.7786296118830858,
|
| 486 |
+
"grad_norm": 17.33100128173828,
|
| 487 |
+
"learning_rate": 1.8259853158708997e-06,
|
| 488 |
+
"loss": 6.0244,
|
| 489 |
+
"step": 650
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.7906085289889794,
|
| 493 |
+
"grad_norm": 10.411093711853027,
|
| 494 |
+
"learning_rate": 1.8206241665754687e-06,
|
| 495 |
+
"loss": 2.8721,
|
| 496 |
+
"step": 660
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 0.802587446094873,
|
| 500 |
+
"grad_norm": 181.8240966796875,
|
| 501 |
+
"learning_rate": 1.815189802910143e-06,
|
| 502 |
+
"loss": 5.1721,
|
| 503 |
+
"step": 670
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 0.8145663632007667,
|
| 507 |
+
"grad_norm": 39.83287048339844,
|
| 508 |
+
"learning_rate": 1.80968270971747e-06,
|
| 509 |
+
"loss": 4.9115,
|
| 510 |
+
"step": 680
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 0.8265452803066603,
|
| 514 |
+
"grad_norm": 39.86928176879883,
|
| 515 |
+
"learning_rate": 1.8041033783287737e-06,
|
| 516 |
+
"loss": 3.8957,
|
| 517 |
+
"step": 690
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 0.8385241974125539,
|
| 521 |
+
"grad_norm": 57.7371711730957,
|
| 522 |
+
"learning_rate": 1.7984523065203188e-06,
|
| 523 |
+
"loss": 3.1863,
|
| 524 |
+
"step": 700
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.8505031145184475,
|
| 528 |
+
"grad_norm": 14.148628234863281,
|
| 529 |
+
"learning_rate": 1.792729998468899e-06,
|
| 530 |
+
"loss": 4.26,
|
| 531 |
+
"step": 710
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.8624820316243411,
|
| 535 |
+
"grad_norm": 71.03279113769531,
|
| 536 |
+
"learning_rate": 1.7869369647068577e-06,
|
| 537 |
+
"loss": 4.9559,
|
| 538 |
+
"step": 720
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.8744609487302348,
|
| 542 |
+
"grad_norm": 17.670730590820312,
|
| 543 |
+
"learning_rate": 1.7810737220765372e-06,
|
| 544 |
+
"loss": 3.9867,
|
| 545 |
+
"step": 730
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 0.8864398658361284,
|
| 549 |
+
"grad_norm": 14.698404312133789,
|
| 550 |
+
"learning_rate": 1.7751407936841684e-06,
|
| 551 |
+
"loss": 2.7134,
|
| 552 |
+
"step": 740
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.898418782942022,
|
| 556 |
+
"grad_norm": 66.42393493652344,
|
| 557 |
+
"learning_rate": 1.7691387088532001e-06,
|
| 558 |
+
"loss": 3.2121,
|
| 559 |
+
"step": 750
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 0.9103977000479156,
|
| 563 |
+
"grad_norm": 76.34748077392578,
|
| 564 |
+
"learning_rate": 1.7630680030770732e-06,
|
| 565 |
+
"loss": 4.7613,
|
| 566 |
+
"step": 760
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 0.9223766171538093,
|
| 570 |
+
"grad_norm": 46.27962112426758,
|
| 571 |
+
"learning_rate": 1.7569292179714465e-06,
|
| 572 |
+
"loss": 3.2976,
|
| 573 |
+
"step": 770
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.934355534259703,
|
| 577 |
+
"grad_norm": 14.20971965789795,
|
| 578 |
+
"learning_rate": 1.750722901225873e-06,
|
| 579 |
+
"loss": 1.9176,
|
| 580 |
+
"step": 780
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 0.9463344513655966,
|
| 584 |
+
"grad_norm": 56.962379455566406,
|
| 585 |
+
"learning_rate": 1.7444496065549384e-06,
|
| 586 |
+
"loss": 1.9859,
|
| 587 |
+
"step": 790
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.9583133684714902,
|
| 591 |
+
"grad_norm": 32.20167541503906,
|
| 592 |
+
"learning_rate": 1.7381098936488574e-06,
|
| 593 |
+
"loss": 6.9549,
|
| 594 |
+
"step": 800
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.9583133684714902,
|
| 598 |
+
"eval_loss": 1.0978227853775024,
|
| 599 |
+
"eval_runtime": 238.5511,
|
| 600 |
+
"eval_samples_per_second": 6.246,
|
| 601 |
+
"eval_steps_per_second": 3.123,
|
| 602 |
+
"step": 800
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.9702922855773838,
|
| 606 |
+
"grad_norm": 82.94165802001953,
|
| 607 |
+
"learning_rate": 1.7317043281235418e-06,
|
| 608 |
+
"loss": 4.1317,
|
| 609 |
+
"step": 810
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.9822712026832774,
|
| 613 |
+
"grad_norm": 110.44422912597656,
|
| 614 |
+
"learning_rate": 1.725233481470135e-06,
|
| 615 |
+
"loss": 3.2924,
|
| 616 |
+
"step": 820
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 0.9942501197891711,
|
| 620 |
+
"grad_norm": 88.77394104003906,
|
| 621 |
+
"learning_rate": 1.7186979310040268e-06,
|
| 622 |
+
"loss": 5.5422,
|
| 623 |
+
"step": 830
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 1.0071873502635362,
|
| 627 |
+
"grad_norm": 167.41412353515625,
|
| 628 |
+
"learning_rate": 1.7120982598133456e-06,
|
| 629 |
+
"loss": 3.5133,
|
| 630 |
+
"step": 840
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 1.0191662673694297,
|
| 634 |
+
"grad_norm": 16.926607131958008,
|
| 635 |
+
"learning_rate": 1.7054350567069364e-06,
|
| 636 |
+
"loss": 4.2376,
|
| 637 |
+
"step": 850
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 1.0311451844753234,
|
| 641 |
+
"grad_norm": 96.12760925292969,
|
| 642 |
+
"learning_rate": 1.698708916161829e-06,
|
| 643 |
+
"loss": 3.6823,
|
| 644 |
+
"step": 860
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 1.0431241015812172,
|
| 648 |
+
"grad_norm": 103.28370666503906,
|
| 649 |
+
"learning_rate": 1.6919204382701987e-06,
|
| 650 |
+
"loss": 2.5705,
|
| 651 |
+
"step": 870
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 1.0551030186871106,
|
| 655 |
+
"grad_norm": 166.0828857421875,
|
| 656 |
+
"learning_rate": 1.6850702286858298e-06,
|
| 657 |
+
"loss": 2.9061,
|
| 658 |
+
"step": 880
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.0670819357930044,
|
| 662 |
+
"grad_norm": 136.93392944335938,
|
| 663 |
+
"learning_rate": 1.678158898570078e-06,
|
| 664 |
+
"loss": 2.8635,
|
| 665 |
+
"step": 890
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.0790608528988979,
|
| 669 |
+
"grad_norm": 14.542271614074707,
|
| 670 |
+
"learning_rate": 1.6711870645373449e-06,
|
| 671 |
+
"loss": 4.2555,
|
| 672 |
+
"step": 900
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.0910397700047916,
|
| 676 |
+
"grad_norm": 53.567359924316406,
|
| 677 |
+
"learning_rate": 1.6641553486000651e-06,
|
| 678 |
+
"loss": 3.1885,
|
| 679 |
+
"step": 910
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.1030186871106853,
|
| 683 |
+
"grad_norm": 100.29656982421875,
|
| 684 |
+
"learning_rate": 1.6570643781132118e-06,
|
| 685 |
+
"loss": 4.953,
|
| 686 |
+
"step": 920
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.1149976042165788,
|
| 690 |
+
"grad_norm": 89.06425476074219,
|
| 691 |
+
"learning_rate": 1.649914785718324e-06,
|
| 692 |
+
"loss": 4.9896,
|
| 693 |
+
"step": 930
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.1269765213224725,
|
| 697 |
+
"grad_norm": 17.858898162841797,
|
| 698 |
+
"learning_rate": 1.6427072092870651e-06,
|
| 699 |
+
"loss": 1.5295,
|
| 700 |
+
"step": 940
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 1.138955438428366,
|
| 704 |
+
"grad_norm": 15.720714569091797,
|
| 705 |
+
"learning_rate": 1.6354422918643133e-06,
|
| 706 |
+
"loss": 3.0117,
|
| 707 |
+
"step": 950
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 1.1509343555342597,
|
| 711 |
+
"grad_norm": 13.404827117919922,
|
| 712 |
+
"learning_rate": 1.628120681610789e-06,
|
| 713 |
+
"loss": 2.1361,
|
| 714 |
+
"step": 960
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.1629132726401532,
|
| 718 |
+
"grad_norm": 130.946044921875,
|
| 719 |
+
"learning_rate": 1.6207430317452297e-06,
|
| 720 |
+
"loss": 3.941,
|
| 721 |
+
"step": 970
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.174892189746047,
|
| 725 |
+
"grad_norm": 14.730375289916992,
|
| 726 |
+
"learning_rate": 1.613310000486108e-06,
|
| 727 |
+
"loss": 3.2318,
|
| 728 |
+
"step": 980
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.1868711068519406,
|
| 732 |
+
"grad_norm": 82.65552520751953,
|
| 733 |
+
"learning_rate": 1.6058222509929096e-06,
|
| 734 |
+
"loss": 3.9045,
|
| 735 |
+
"step": 990
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 1.1988500239578341,
|
| 739 |
+
"grad_norm": Infinity,
|
| 740 |
+
"learning_rate": 1.5982804513069664e-06,
|
| 741 |
+
"loss": 5.5404,
|
| 742 |
+
"step": 1000
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 1.1988500239578341,
|
| 746 |
+
"eval_loss": 1.1037527322769165,
|
| 747 |
+
"eval_runtime": 238.5531,
|
| 748 |
+
"eval_samples_per_second": 6.246,
|
| 749 |
+
"eval_steps_per_second": 3.123,
|
| 750 |
+
"step": 1000
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 1.2108289410637278,
|
| 754 |
+
"grad_norm": 67.75409698486328,
|
| 755 |
+
"learning_rate": 1.5914471746978935e-06,
|
| 756 |
+
"loss": 2.6392,
|
| 757 |
+
"step": 1010
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 1.2228078581696216,
|
| 761 |
+
"grad_norm": 49.21822738647461,
|
| 762 |
+
"learning_rate": 1.5838045373221053e-06,
|
| 763 |
+
"loss": 4.0259,
|
| 764 |
+
"step": 1020
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 1.234786775275515,
|
| 768 |
+
"grad_norm": 228.187744140625,
|
| 769 |
+
"learning_rate": 1.5761098141278849e-06,
|
| 770 |
+
"loss": 5.8343,
|
| 771 |
+
"step": 1030
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 1.2467656923814088,
|
| 775 |
+
"grad_norm": 33.68708419799805,
|
| 776 |
+
"learning_rate": 1.5683636916223236e-06,
|
| 777 |
+
"loss": 3.9807,
|
| 778 |
+
"step": 1040
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 1.2587446094873023,
|
| 782 |
+
"grad_norm": 140.39187622070312,
|
| 783 |
+
"learning_rate": 1.5605668608982526e-06,
|
| 784 |
+
"loss": 3.9716,
|
| 785 |
+
"step": 1050
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 1.270723526593196,
|
| 789 |
+
"grad_norm": 11.902155876159668,
|
| 790 |
+
"learning_rate": 1.5527200175725842e-06,
|
| 791 |
+
"loss": 3.2315,
|
| 792 |
+
"step": 1060
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 1.2827024436990895,
|
| 796 |
+
"grad_norm": 144.3499298095703,
|
| 797 |
+
"learning_rate": 1.5448238617242488e-06,
|
| 798 |
+
"loss": 2.6336,
|
| 799 |
+
"step": 1070
|
| 800 |
+
},
|
| 801 |
+
{
|
| 802 |
+
"epoch": 1.2946813608049832,
|
| 803 |
+
"grad_norm": 59.747928619384766,
|
| 804 |
+
"learning_rate": 1.5368790978317395e-06,
|
| 805 |
+
"loss": 3.206,
|
| 806 |
+
"step": 1080
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 1.306660277910877,
|
| 810 |
+
"grad_norm": 53.534950256347656,
|
| 811 |
+
"learning_rate": 1.5288864347102545e-06,
|
| 812 |
+
"loss": 4.3036,
|
| 813 |
+
"step": 1090
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 1.3186391950167704,
|
| 817 |
+
"grad_norm": 48.62434387207031,
|
| 818 |
+
"learning_rate": 1.520846585448463e-06,
|
| 819 |
+
"loss": 2.4486,
|
| 820 |
+
"step": 1100
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 1.3306181121226641,
|
| 824 |
+
"grad_norm": 18.836971282958984,
|
| 825 |
+
"learning_rate": 1.512760267344882e-06,
|
| 826 |
+
"loss": 4.0121,
|
| 827 |
+
"step": 1110
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 1.3425970292285578,
|
| 831 |
+
"grad_norm": 224.24917602539062,
|
| 832 |
+
"learning_rate": 1.5046282018438814e-06,
|
| 833 |
+
"loss": 2.8545,
|
| 834 |
+
"step": 1120
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 1.3545759463344513,
|
| 838 |
+
"grad_norm": 18.120187759399414,
|
| 839 |
+
"learning_rate": 1.4964511144713174e-06,
|
| 840 |
+
"loss": 3.1619,
|
| 841 |
+
"step": 1130
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 1.366554863440345,
|
| 845 |
+
"grad_norm": 39.76359939575195,
|
| 846 |
+
"learning_rate": 1.4882297347698048e-06,
|
| 847 |
+
"loss": 3.0413,
|
| 848 |
+
"step": 1140
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 1.3785337805462385,
|
| 852 |
+
"grad_norm": 64.61255645751953,
|
| 853 |
+
"learning_rate": 1.4799647962336255e-06,
|
| 854 |
+
"loss": 3.8001,
|
| 855 |
+
"step": 1150
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 1.3905126976521323,
|
| 859 |
+
"grad_norm": 12.653026580810547,
|
| 860 |
+
"learning_rate": 1.471657036243291e-06,
|
| 861 |
+
"loss": 5.532,
|
| 862 |
+
"step": 1160
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 1.4024916147580258,
|
| 866 |
+
"grad_norm": 53.37699508666992,
|
| 867 |
+
"learning_rate": 1.4633071959997525e-06,
|
| 868 |
+
"loss": 3.4156,
|
| 869 |
+
"step": 1170
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 1.4144705318639195,
|
| 873 |
+
"grad_norm": 37.08938217163086,
|
| 874 |
+
"learning_rate": 1.4549160204582731e-06,
|
| 875 |
+
"loss": 2.5073,
|
| 876 |
+
"step": 1180
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 1.4264494489698132,
|
| 880 |
+
"grad_norm": 139.479736328125,
|
| 881 |
+
"learning_rate": 1.4464842582619652e-06,
|
| 882 |
+
"loss": 3.36,
|
| 883 |
+
"step": 1190
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 1.4384283660757067,
|
| 887 |
+
"grad_norm": 88.59599304199219,
|
| 888 |
+
"learning_rate": 1.4380126616749975e-06,
|
| 889 |
+
"loss": 5.2213,
|
| 890 |
+
"step": 1200
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.4384283660757067,
|
| 894 |
+
"eval_loss": 1.1036500930786133,
|
| 895 |
+
"eval_runtime": 238.313,
|
| 896 |
+
"eval_samples_per_second": 6.252,
|
| 897 |
+
"eval_steps_per_second": 3.126,
|
| 898 |
+
"step": 1200
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 1.4504072831816004,
|
| 902 |
+
"grad_norm": 129.61988830566406,
|
| 903 |
+
"learning_rate": 1.4295019865154785e-06,
|
| 904 |
+
"loss": 7.1682,
|
| 905 |
+
"step": 1210
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 1.462386200287494,
|
| 909 |
+
"grad_norm": 19.823545455932617,
|
| 910 |
+
"learning_rate": 1.4209529920880272e-06,
|
| 911 |
+
"loss": 4.6843,
|
| 912 |
+
"step": 1220
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.4743651173933876,
|
| 916 |
+
"grad_norm": 107.53630065917969,
|
| 917 |
+
"learning_rate": 1.4123664411160252e-06,
|
| 918 |
+
"loss": 2.4525,
|
| 919 |
+
"step": 1230
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 1.4863440344992813,
|
| 923 |
+
"grad_norm": 54.906280517578125,
|
| 924 |
+
"learning_rate": 1.4037430996735722e-06,
|
| 925 |
+
"loss": 5.9388,
|
| 926 |
+
"step": 1240
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 1.4983229516051748,
|
| 930 |
+
"grad_norm": 77.29488372802734,
|
| 931 |
+
"learning_rate": 1.3950837371171355e-06,
|
| 932 |
+
"loss": 5.3705,
|
| 933 |
+
"step": 1250
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 1.5103018687110685,
|
| 937 |
+
"grad_norm": 115.9428482055664,
|
| 938 |
+
"learning_rate": 1.3863891260169114e-06,
|
| 939 |
+
"loss": 4.0317,
|
| 940 |
+
"step": 1260
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 1.522280785816962,
|
| 944 |
+
"grad_norm": 19.207189559936523,
|
| 945 |
+
"learning_rate": 1.3776600420878973e-06,
|
| 946 |
+
"loss": 3.8767,
|
| 947 |
+
"step": 1270
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 1.5342597029228557,
|
| 951 |
+
"grad_norm": 83.13814544677734,
|
| 952 |
+
"learning_rate": 1.3688972641206837e-06,
|
| 953 |
+
"loss": 4.5835,
|
| 954 |
+
"step": 1280
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 1.5462386200287495,
|
| 958 |
+
"grad_norm": 222.7005157470703,
|
| 959 |
+
"learning_rate": 1.3601015739119733e-06,
|
| 960 |
+
"loss": 3.3379,
|
| 961 |
+
"step": 1290
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 1.558217537134643,
|
| 965 |
+
"grad_norm": 51.51054382324219,
|
| 966 |
+
"learning_rate": 1.35127375619483e-06,
|
| 967 |
+
"loss": 5.4397,
|
| 968 |
+
"step": 1300
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 1.5701964542405367,
|
| 972 |
+
"grad_norm": 109.09092712402344,
|
| 973 |
+
"learning_rate": 1.3424145985686662e-06,
|
| 974 |
+
"loss": 3.1896,
|
| 975 |
+
"step": 1310
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 1.5821753713464304,
|
| 979 |
+
"grad_norm": 11.890337944030762,
|
| 980 |
+
"learning_rate": 1.333524891428976e-06,
|
| 981 |
+
"loss": 4.4828,
|
| 982 |
+
"step": 1320
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 1.5941542884523239,
|
| 986 |
+
"grad_norm": 18.87173843383789,
|
| 987 |
+
"learning_rate": 1.324605427896817e-06,
|
| 988 |
+
"loss": 2.4719,
|
| 989 |
+
"step": 1330
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 1.6061332055582176,
|
| 993 |
+
"grad_norm": 110.26778411865234,
|
| 994 |
+
"learning_rate": 1.3156570037480497e-06,
|
| 995 |
+
"loss": 3.4721,
|
| 996 |
+
"step": 1340
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 1.6181121226641113,
|
| 1000 |
+
"grad_norm": 19.194913864135742,
|
| 1001 |
+
"learning_rate": 1.3066804173423397e-06,
|
| 1002 |
+
"loss": 4.3532,
|
| 1003 |
+
"step": 1350
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 1.6300910397700048,
|
| 1007 |
+
"grad_norm": 89.65567016601562,
|
| 1008 |
+
"learning_rate": 1.297676469551931e-06,
|
| 1009 |
+
"loss": 4.1742,
|
| 1010 |
+
"step": 1360
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 1.6420699568758983,
|
| 1014 |
+
"grad_norm": 15.179718017578125,
|
| 1015 |
+
"learning_rate": 1.2886459636901927e-06,
|
| 1016 |
+
"loss": 4.2612,
|
| 1017 |
+
"step": 1370
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 1.654048873981792,
|
| 1021 |
+
"grad_norm": 14.557687759399414,
|
| 1022 |
+
"learning_rate": 1.2795897054399498e-06,
|
| 1023 |
+
"loss": 5.2594,
|
| 1024 |
+
"step": 1380
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 1.6660277910876857,
|
| 1028 |
+
"grad_norm": 20.148624420166016,
|
| 1029 |
+
"learning_rate": 1.2705085027816008e-06,
|
| 1030 |
+
"loss": 3.3919,
|
| 1031 |
+
"step": 1390
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 1.6780067081935792,
|
| 1035 |
+
"grad_norm": 9.0452241897583,
|
| 1036 |
+
"learning_rate": 1.261403165921032e-06,
|
| 1037 |
+
"loss": 4.3208,
|
| 1038 |
+
"step": 1400
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 1.6780067081935792,
|
| 1042 |
+
"eval_loss": 1.0974289178848267,
|
| 1043 |
+
"eval_runtime": 238.4129,
|
| 1044 |
+
"eval_samples_per_second": 6.25,
|
| 1045 |
+
"eval_steps_per_second": 3.125,
|
| 1046 |
+
"step": 1400
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 1.689985625299473,
|
| 1050 |
+
"grad_norm": 18.7681941986084,
|
| 1051 |
+
"learning_rate": 1.2522745072173336e-06,
|
| 1052 |
+
"loss": 2.5784,
|
| 1053 |
+
"step": 1410
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 1.7019645424053667,
|
| 1057 |
+
"grad_norm": 112.4189224243164,
|
| 1058 |
+
"learning_rate": 1.243123341110321e-06,
|
| 1059 |
+
"loss": 4.0173,
|
| 1060 |
+
"step": 1420
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 1.7139434595112601,
|
| 1064 |
+
"grad_norm": 13.051095008850098,
|
| 1065 |
+
"learning_rate": 1.2339504840478738e-06,
|
| 1066 |
+
"loss": 3.1098,
|
| 1067 |
+
"step": 1430
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 1.7259223766171539,
|
| 1071 |
+
"grad_norm": 16.918392181396484,
|
| 1072 |
+
"learning_rate": 1.224756754413092e-06,
|
| 1073 |
+
"loss": 3.1983,
|
| 1074 |
+
"step": 1440
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 1.7379012937230476,
|
| 1078 |
+
"grad_norm": 167.8545379638672,
|
| 1079 |
+
"learning_rate": 1.2155429724512838e-06,
|
| 1080 |
+
"loss": 4.8368,
|
| 1081 |
+
"step": 1450
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 1.749880210828941,
|
| 1085 |
+
"grad_norm": 19.237834930419922,
|
| 1086 |
+
"learning_rate": 1.206309960196784e-06,
|
| 1087 |
+
"loss": 3.1809,
|
| 1088 |
+
"step": 1460
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 1.7618591279348346,
|
| 1092 |
+
"grad_norm": 95.0063247680664,
|
| 1093 |
+
"learning_rate": 1.1970585413996132e-06,
|
| 1094 |
+
"loss": 3.9006,
|
| 1095 |
+
"step": 1470
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 1.7738380450407283,
|
| 1099 |
+
"grad_norm": 55.70530319213867,
|
| 1100 |
+
"learning_rate": 1.1877895414519858e-06,
|
| 1101 |
+
"loss": 3.3394,
|
| 1102 |
+
"step": 1480
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 1.785816962146622,
|
| 1106 |
+
"grad_norm": 23.173871994018555,
|
| 1107 |
+
"learning_rate": 1.1785037873146697e-06,
|
| 1108 |
+
"loss": 2.4079,
|
| 1109 |
+
"step": 1490
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 1.7977958792525155,
|
| 1113 |
+
"grad_norm": 120.7456283569336,
|
| 1114 |
+
"learning_rate": 1.1692021074432054e-06,
|
| 1115 |
+
"loss": 4.2111,
|
| 1116 |
+
"step": 1500
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 1.8097747963584092,
|
| 1120 |
+
"grad_norm": 89.36892700195312,
|
| 1121 |
+
"learning_rate": 1.1598853317139958e-06,
|
| 1122 |
+
"loss": 1.8205,
|
| 1123 |
+
"step": 1510
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 1.821753713464303,
|
| 1127 |
+
"grad_norm": 66.96819305419922,
|
| 1128 |
+
"learning_rate": 1.150554291350263e-06,
|
| 1129 |
+
"loss": 4.6707,
|
| 1130 |
+
"step": 1520
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 1.8337326305701964,
|
| 1134 |
+
"grad_norm": 52.6048583984375,
|
| 1135 |
+
"learning_rate": 1.1412098188478914e-06,
|
| 1136 |
+
"loss": 2.2611,
|
| 1137 |
+
"step": 1530
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 1.8457115476760901,
|
| 1141 |
+
"grad_norm": 100.39757537841797,
|
| 1142 |
+
"learning_rate": 1.1318527479011513e-06,
|
| 1143 |
+
"loss": 3.3554,
|
| 1144 |
+
"step": 1540
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 1.8576904647819839,
|
| 1148 |
+
"grad_norm": 12.364606857299805,
|
| 1149 |
+
"learning_rate": 1.1224839133283208e-06,
|
| 1150 |
+
"loss": 2.7868,
|
| 1151 |
+
"step": 1550
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 1.8696693818878773,
|
| 1155 |
+
"grad_norm": 93.31403350830078,
|
| 1156 |
+
"learning_rate": 1.1131041509972032e-06,
|
| 1157 |
+
"loss": 3.7607,
|
| 1158 |
+
"step": 1560
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 1.8816482989937708,
|
| 1162 |
+
"grad_norm": 77.30158233642578,
|
| 1163 |
+
"learning_rate": 1.1037142977505548e-06,
|
| 1164 |
+
"loss": 3.28,
|
| 1165 |
+
"step": 1570
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 1.8936272160996646,
|
| 1169 |
+
"grad_norm": 13.82420825958252,
|
| 1170 |
+
"learning_rate": 1.0943151913314211e-06,
|
| 1171 |
+
"loss": 3.3544,
|
| 1172 |
+
"step": 1580
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 1.9056061332055583,
|
| 1176 |
+
"grad_norm": 16.398128509521484,
|
| 1177 |
+
"learning_rate": 1.084907670308397e-06,
|
| 1178 |
+
"loss": 2.7871,
|
| 1179 |
+
"step": 1590
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 1.9175850503114518,
|
| 1183 |
+
"grad_norm": 19.750925064086914,
|
| 1184 |
+
"learning_rate": 1.0754925740008098e-06,
|
| 1185 |
+
"loss": 4.1985,
|
| 1186 |
+
"step": 1600
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 1.9175850503114518,
|
| 1190 |
+
"eval_loss": 1.1042989492416382,
|
| 1191 |
+
"eval_runtime": 238.4482,
|
| 1192 |
+
"eval_samples_per_second": 6.249,
|
| 1193 |
+
"eval_steps_per_second": 3.124,
|
| 1194 |
+
"step": 1600
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 1.9295639674173455,
|
| 1198 |
+
"grad_norm": 16.436159133911133,
|
| 1199 |
+
"learning_rate": 1.066070742403839e-06,
|
| 1200 |
+
"loss": 3.9566,
|
| 1201 |
+
"step": 1610
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 1.9415428845232392,
|
| 1205 |
+
"grad_norm": 67.2239761352539,
|
| 1206 |
+
"learning_rate": 1.056643016113572e-06,
|
| 1207 |
+
"loss": 4.0604,
|
| 1208 |
+
"step": 1620
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 1.9535218016291327,
|
| 1212 |
+
"grad_norm": 52.419456481933594,
|
| 1213 |
+
"learning_rate": 1.047210236252008e-06,
|
| 1214 |
+
"loss": 4.4566,
|
| 1215 |
+
"step": 1630
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 1.9655007187350264,
|
| 1219 |
+
"grad_norm": 102.50648498535156,
|
| 1220 |
+
"learning_rate": 1.0377732443920155e-06,
|
| 1221 |
+
"loss": 2.5929,
|
| 1222 |
+
"step": 1640
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 1.9774796358409201,
|
| 1226 |
+
"grad_norm": 85.2761459350586,
|
| 1227 |
+
"learning_rate": 1.0283328824822498e-06,
|
| 1228 |
+
"loss": 3.278,
|
| 1229 |
+
"step": 1650
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 1.9894585529468136,
|
| 1233 |
+
"grad_norm": 111.09942626953125,
|
| 1234 |
+
"learning_rate": 1.0188899927720324e-06,
|
| 1235 |
+
"loss": 2.1727,
|
| 1236 |
+
"step": 1660
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 2.0023957834211785,
|
| 1240 |
+
"grad_norm": 88.2173080444336,
|
| 1241 |
+
"learning_rate": 1.009445417736213e-06,
|
| 1242 |
+
"loss": 4.7098,
|
| 1243 |
+
"step": 1670
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 2.0143747005270725,
|
| 1247 |
+
"grad_norm": 15.37248420715332,
|
| 1248 |
+
"learning_rate": 1e-06,
|
| 1249 |
+
"loss": 3.3523,
|
| 1250 |
+
"step": 1680
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 2.026353617632966,
|
| 1254 |
+
"grad_norm": 428.4352111816406,
|
| 1255 |
+
"learning_rate": 9.905545822637871e-07,
|
| 1256 |
+
"loss": 4.4776,
|
| 1257 |
+
"step": 1690
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 2.0383325347388594,
|
| 1261 |
+
"grad_norm": 16.143062591552734,
|
| 1262 |
+
"learning_rate": 9.811100072279673e-07,
|
| 1263 |
+
"loss": 3.2249,
|
| 1264 |
+
"step": 1700
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"epoch": 2.0503114518447534,
|
| 1268 |
+
"grad_norm": 142.73899841308594,
|
| 1269 |
+
"learning_rate": 9.716671175177506e-07,
|
| 1270 |
+
"loss": 3.6488,
|
| 1271 |
+
"step": 1710
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 2.062290368950647,
|
| 1275 |
+
"grad_norm": 86.692626953125,
|
| 1276 |
+
"learning_rate": 9.622267556079844e-07,
|
| 1277 |
+
"loss": 2.3491,
|
| 1278 |
+
"step": 1720
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 2.0742692860565404,
|
| 1282 |
+
"grad_norm": 127.98802947998047,
|
| 1283 |
+
"learning_rate": 9.527897637479921e-07,
|
| 1284 |
+
"loss": 4.828,
|
| 1285 |
+
"step": 1730
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 2.0862482031624343,
|
| 1289 |
+
"grad_norm": 14.027091026306152,
|
| 1290 |
+
"learning_rate": 9.433569838864282e-07,
|
| 1291 |
+
"loss": 4.176,
|
| 1292 |
+
"step": 1740
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 2.098227120268328,
|
| 1296 |
+
"grad_norm": 16.68279266357422,
|
| 1297 |
+
"learning_rate": 9.33929257596161e-07,
|
| 1298 |
+
"loss": 2.8037,
|
| 1299 |
+
"step": 1750
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 2.1102060373742213,
|
| 1303 |
+
"grad_norm": 68.17903900146484,
|
| 1304 |
+
"learning_rate": 9.245074259991904e-07,
|
| 1305 |
+
"loss": 3.448,
|
| 1306 |
+
"step": 1760
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 2.122184954480115,
|
| 1310 |
+
"grad_norm": 90.32327270507812,
|
| 1311 |
+
"learning_rate": 9.150923296916032e-07,
|
| 1312 |
+
"loss": 3.3154,
|
| 1313 |
+
"step": 1770
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 2.1341638715860087,
|
| 1317 |
+
"grad_norm": 218.0877227783203,
|
| 1318 |
+
"learning_rate": 9.056848086685789e-07,
|
| 1319 |
+
"loss": 3.8818,
|
| 1320 |
+
"step": 1780
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 2.1461427886919022,
|
| 1324 |
+
"grad_norm": 117.9521484375,
|
| 1325 |
+
"learning_rate": 8.96285702249445e-07,
|
| 1326 |
+
"loss": 3.7345,
|
| 1327 |
+
"step": 1790
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 2.1581217057977957,
|
| 1331 |
+
"grad_norm": 14.993733406066895,
|
| 1332 |
+
"learning_rate": 8.868958490027966e-07,
|
| 1333 |
+
"loss": 3.3709,
|
| 1334 |
+
"step": 1800
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 2.1581217057977957,
|
| 1338 |
+
"eval_loss": 1.1172066926956177,
|
| 1339 |
+
"eval_runtime": 238.6675,
|
| 1340 |
+
"eval_samples_per_second": 6.243,
|
| 1341 |
+
"eval_steps_per_second": 3.121,
|
| 1342 |
+
"step": 1800
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 2.1701006229036897,
|
| 1346 |
+
"grad_norm": 273.74365234375,
|
| 1347 |
+
"learning_rate": 8.775160866716791e-07,
|
| 1348 |
+
"loss": 4.06,
|
| 1349 |
+
"step": 1810
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 2.182079540009583,
|
| 1353 |
+
"grad_norm": 11.733575820922852,
|
| 1354 |
+
"learning_rate": 8.681472520988488e-07,
|
| 1355 |
+
"loss": 2.4028,
|
| 1356 |
+
"step": 1820
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 2.1940584571154766,
|
| 1360 |
+
"grad_norm": 270.543212890625,
|
| 1361 |
+
"learning_rate": 8.587901811521087e-07,
|
| 1362 |
+
"loss": 5.1853,
|
| 1363 |
+
"step": 1830
|
| 1364 |
+
},
|
| 1365 |
+
{
|
| 1366 |
+
"epoch": 2.2060373742213706,
|
| 1367 |
+
"grad_norm": 18.38782501220703,
|
| 1368 |
+
"learning_rate": 8.494457086497368e-07,
|
| 1369 |
+
"loss": 1.9458,
|
| 1370 |
+
"step": 1840
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 2.218016291327264,
|
| 1374 |
+
"grad_norm": 127.68826293945312,
|
| 1375 |
+
"learning_rate": 8.401146682860041e-07,
|
| 1376 |
+
"loss": 3.2338,
|
| 1377 |
+
"step": 1850
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 2.2299952084331576,
|
| 1381 |
+
"grad_norm": 186.4503173828125,
|
| 1382 |
+
"learning_rate": 8.307978925567945e-07,
|
| 1383 |
+
"loss": 2.5094,
|
| 1384 |
+
"step": 1860
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 2.241974125539051,
|
| 1388 |
+
"grad_norm": 15.038443565368652,
|
| 1389 |
+
"learning_rate": 8.214962126853307e-07,
|
| 1390 |
+
"loss": 2.6388,
|
| 1391 |
+
"step": 1870
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 2.253953042644945,
|
| 1395 |
+
"grad_norm": 74.26438903808594,
|
| 1396 |
+
"learning_rate": 8.122104585480143e-07,
|
| 1397 |
+
"loss": 2.24,
|
| 1398 |
+
"step": 1880
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 2.2659319597508385,
|
| 1402 |
+
"grad_norm": 57.38971710205078,
|
| 1403 |
+
"learning_rate": 8.029414586003866e-07,
|
| 1404 |
+
"loss": 4.3915,
|
| 1405 |
+
"step": 1890
|
| 1406 |
+
},
|
| 1407 |
+
{
|
| 1408 |
+
"epoch": 2.277910876856732,
|
| 1409 |
+
"grad_norm": 221.86866760253906,
|
| 1410 |
+
"learning_rate": 7.93690039803216e-07,
|
| 1411 |
+
"loss": 3.0979,
|
| 1412 |
+
"step": 1900
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 2.289889793962626,
|
| 1416 |
+
"grad_norm": 66.43515014648438,
|
| 1417 |
+
"learning_rate": 7.844570275487159e-07,
|
| 1418 |
+
"loss": 2.0459,
|
| 1419 |
+
"step": 1910
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 2.3018687110685194,
|
| 1423 |
+
"grad_norm": 174.6913604736328,
|
| 1424 |
+
"learning_rate": 7.752432455869081e-07,
|
| 1425 |
+
"loss": 4.3113,
|
| 1426 |
+
"step": 1920
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 2.313847628174413,
|
| 1430 |
+
"grad_norm": 12.269379615783691,
|
| 1431 |
+
"learning_rate": 7.660495159521264e-07,
|
| 1432 |
+
"loss": 2.6802,
|
| 1433 |
+
"step": 1930
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 2.3258265452803064,
|
| 1437 |
+
"grad_norm": 14.982316970825195,
|
| 1438 |
+
"learning_rate": 7.56876658889679e-07,
|
| 1439 |
+
"loss": 3.2325,
|
| 1440 |
+
"step": 1940
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 2.3378054623862004,
|
| 1444 |
+
"grad_norm": 75.86939239501953,
|
| 1445 |
+
"learning_rate": 7.477254927826664e-07,
|
| 1446 |
+
"loss": 1.0064,
|
| 1447 |
+
"step": 1950
|
| 1448 |
+
},
|
| 1449 |
+
{
|
| 1450 |
+
"epoch": 2.349784379492094,
|
| 1451 |
+
"grad_norm": 12.59216594696045,
|
| 1452 |
+
"learning_rate": 7.38596834078968e-07,
|
| 1453 |
+
"loss": 2.3464,
|
| 1454 |
+
"step": 1960
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 2.361763296597988,
|
| 1458 |
+
"grad_norm": 44.80192184448242,
|
| 1459 |
+
"learning_rate": 7.294914972183992e-07,
|
| 1460 |
+
"loss": 3.9336,
|
| 1461 |
+
"step": 1970
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 2.3737422137038813,
|
| 1465 |
+
"grad_norm": 227.7502899169922,
|
| 1466 |
+
"learning_rate": 7.204102945600502e-07,
|
| 1467 |
+
"loss": 3.3652,
|
| 1468 |
+
"step": 1980
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 2.3857211308097748,
|
| 1472 |
+
"grad_norm": 202.1083526611328,
|
| 1473 |
+
"learning_rate": 7.113540363098072e-07,
|
| 1474 |
+
"loss": 3.0293,
|
| 1475 |
+
"step": 1990
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 2.3977000479156683,
|
| 1479 |
+
"grad_norm": 148.12112426757812,
|
| 1480 |
+
"learning_rate": 7.02323530448069e-07,
|
| 1481 |
+
"loss": 3.7548,
|
| 1482 |
+
"step": 2000
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 2.3977000479156683,
|
| 1486 |
+
"eval_loss": 1.1212018728256226,
|
| 1487 |
+
"eval_runtime": 237.7587,
|
| 1488 |
+
"eval_samples_per_second": 6.267,
|
| 1489 |
+
"eval_steps_per_second": 3.133,
|
| 1490 |
+
"step": 2000
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 2.409678965021562,
|
| 1494 |
+
"grad_norm": 107.33174133300781,
|
| 1495 |
+
"learning_rate": 6.933195826576603e-07,
|
| 1496 |
+
"loss": 3.9499,
|
| 1497 |
+
"step": 2010
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 2.4216578821274557,
|
| 1501 |
+
"grad_norm": 15.232194900512695,
|
| 1502 |
+
"learning_rate": 6.843429962519504e-07,
|
| 1503 |
+
"loss": 4.3203,
|
| 1504 |
+
"step": 2020
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"epoch": 2.433636799233349,
|
| 1508 |
+
"grad_norm": 14.64511489868164,
|
| 1509 |
+
"learning_rate": 6.75394572103183e-07,
|
| 1510 |
+
"loss": 4.5243,
|
| 1511 |
+
"step": 2030
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 2.445615716339243,
|
| 1515 |
+
"grad_norm": 16.1613712310791,
|
| 1516 |
+
"learning_rate": 6.664751085710239e-07,
|
| 1517 |
+
"loss": 3.5644,
|
| 1518 |
+
"step": 2040
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 2.4575946334451366,
|
| 1522 |
+
"grad_norm": 61.8858642578125,
|
| 1523 |
+
"learning_rate": 6.575854014313338e-07,
|
| 1524 |
+
"loss": 3.7972,
|
| 1525 |
+
"step": 2050
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 2.46957355055103,
|
| 1529 |
+
"grad_norm": 52.300514221191406,
|
| 1530 |
+
"learning_rate": 6.487262438051701e-07,
|
| 1531 |
+
"loss": 3.6956,
|
| 1532 |
+
"step": 2060
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 2.4815524676569236,
|
| 1536 |
+
"grad_norm": 208.3380126953125,
|
| 1537 |
+
"learning_rate": 6.398984260880266e-07,
|
| 1538 |
+
"loss": 3.6895,
|
| 1539 |
+
"step": 2070
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 2.4935313847628175,
|
| 1543 |
+
"grad_norm": 18.773216247558594,
|
| 1544 |
+
"learning_rate": 6.311027358793166e-07,
|
| 1545 |
+
"loss": 3.0383,
|
| 1546 |
+
"step": 2080
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 2.505510301868711,
|
| 1550 |
+
"grad_norm": 19.703941345214844,
|
| 1551 |
+
"learning_rate": 6.223399579121029e-07,
|
| 1552 |
+
"loss": 2.5712,
|
| 1553 |
+
"step": 2090
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 2.5174892189746045,
|
| 1557 |
+
"grad_norm": 63.63347625732422,
|
| 1558 |
+
"learning_rate": 6.136108739830886e-07,
|
| 1559 |
+
"loss": 2.2939,
|
| 1560 |
+
"step": 2100
|
| 1561 |
+
},
|
| 1562 |
+
{
|
| 1563 |
+
"epoch": 2.5294681360804985,
|
| 1564 |
+
"grad_norm": 23.118276596069336,
|
| 1565 |
+
"learning_rate": 6.049162628828644e-07,
|
| 1566 |
+
"loss": 3.329,
|
| 1567 |
+
"step": 2110
|
| 1568 |
+
},
|
| 1569 |
+
{
|
| 1570 |
+
"epoch": 2.541447053186392,
|
| 1571 |
+
"grad_norm": 16.502140045166016,
|
| 1572 |
+
"learning_rate": 5.962569003264276e-07,
|
| 1573 |
+
"loss": 4.0458,
|
| 1574 |
+
"step": 2120
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 2.5534259702922855,
|
| 1578 |
+
"grad_norm": 11.94331169128418,
|
| 1579 |
+
"learning_rate": 5.876335588839746e-07,
|
| 1580 |
+
"loss": 3.7107,
|
| 1581 |
+
"step": 2130
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 2.565404887398179,
|
| 1585 |
+
"grad_norm": 20.692808151245117,
|
| 1586 |
+
"learning_rate": 5.79047007911973e-07,
|
| 1587 |
+
"loss": 2.3799,
|
| 1588 |
+
"step": 2140
|
| 1589 |
+
},
|
| 1590 |
+
{
|
| 1591 |
+
"epoch": 2.577383804504073,
|
| 1592 |
+
"grad_norm": 170.5503692626953,
|
| 1593 |
+
"learning_rate": 5.704980134845213e-07,
|
| 1594 |
+
"loss": 2.808,
|
| 1595 |
+
"step": 2150
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 2.5893627216099664,
|
| 1599 |
+
"grad_norm": 17.668119430541992,
|
| 1600 |
+
"learning_rate": 5.619873383250029e-07,
|
| 1601 |
+
"loss": 2.4657,
|
| 1602 |
+
"step": 2160
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 2.6013416387158603,
|
| 1606 |
+
"grad_norm": 12.804652214050293,
|
| 1607 |
+
"learning_rate": 5.535157417380346e-07,
|
| 1608 |
+
"loss": 4.2857,
|
| 1609 |
+
"step": 2170
|
| 1610 |
+
},
|
| 1611 |
+
{
|
| 1612 |
+
"epoch": 2.613320555821754,
|
| 1613 |
+
"grad_norm": 231.0024871826172,
|
| 1614 |
+
"learning_rate": 5.450839795417266e-07,
|
| 1615 |
+
"loss": 5.443,
|
| 1616 |
+
"step": 2180
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 2.6252994729276473,
|
| 1620 |
+
"grad_norm": 156.8734893798828,
|
| 1621 |
+
"learning_rate": 5.366928040002476e-07,
|
| 1622 |
+
"loss": 4.17,
|
| 1623 |
+
"step": 2190
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 2.637278390033541,
|
| 1627 |
+
"grad_norm": 49.93342590332031,
|
| 1628 |
+
"learning_rate": 5.283429637567091e-07,
|
| 1629 |
+
"loss": 3.2694,
|
| 1630 |
+
"step": 2200
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 2.637278390033541,
|
| 1634 |
+
"eval_loss": 1.1175537109375,
|
| 1635 |
+
"eval_runtime": 238.0084,
|
| 1636 |
+
"eval_samples_per_second": 6.26,
|
| 1637 |
+
"eval_steps_per_second": 3.13,
|
| 1638 |
+
"step": 2200
|
| 1639 |
+
},
|
| 1640 |
+
{
|
| 1641 |
+
"epoch": 2.6492573071394347,
|
| 1642 |
+
"grad_norm": 11.659183502197266,
|
| 1643 |
+
"learning_rate": 5.200352037663745e-07,
|
| 1644 |
+
"loss": 1.2186,
|
| 1645 |
+
"step": 2210
|
| 1646 |
+
},
|
| 1647 |
+
{
|
| 1648 |
+
"epoch": 2.6612362242453282,
|
| 1649 |
+
"grad_norm": 16.97669219970703,
|
| 1650 |
+
"learning_rate": 5.117702652301952e-07,
|
| 1651 |
+
"loss": 3.9984,
|
| 1652 |
+
"step": 2220
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 2.6732151413512217,
|
| 1656 |
+
"grad_norm": 49.165199279785156,
|
| 1657 |
+
"learning_rate": 5.035488855286823e-07,
|
| 1658 |
+
"loss": 2.916,
|
| 1659 |
+
"step": 2230
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 2.6851940584571157,
|
| 1663 |
+
"grad_norm": 14.636496543884277,
|
| 1664 |
+
"learning_rate": 4.953717981561186e-07,
|
| 1665 |
+
"loss": 3.1995,
|
| 1666 |
+
"step": 2240
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 2.697172975563009,
|
| 1670 |
+
"grad_norm": 13.993462562561035,
|
| 1671 |
+
"learning_rate": 4.872397326551179e-07,
|
| 1672 |
+
"loss": 3.1531,
|
| 1673 |
+
"step": 2250
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 2.7091518926689027,
|
| 1677 |
+
"grad_norm": 14.866923332214355,
|
| 1678 |
+
"learning_rate": 4.791534145515368e-07,
|
| 1679 |
+
"loss": 2.7951,
|
| 1680 |
+
"step": 2260
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 2.721130809774796,
|
| 1684 |
+
"grad_norm": 12.874122619628906,
|
| 1685 |
+
"learning_rate": 4.711135652897452e-07,
|
| 1686 |
+
"loss": 4.0197,
|
| 1687 |
+
"step": 2270
|
| 1688 |
+
},
|
| 1689 |
+
{
|
| 1690 |
+
"epoch": 2.73310972688069,
|
| 1691 |
+
"grad_norm": 166.74929809570312,
|
| 1692 |
+
"learning_rate": 4.6312090216826074e-07,
|
| 1693 |
+
"loss": 4.324,
|
| 1694 |
+
"step": 2280
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 2.7450886439865836,
|
| 1698 |
+
"grad_norm": 169.29539489746094,
|
| 1699 |
+
"learning_rate": 4.551761382757513e-07,
|
| 1700 |
+
"loss": 4.1737,
|
| 1701 |
+
"step": 2290
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 2.757067561092477,
|
| 1705 |
+
"grad_norm": 12.698741912841797,
|
| 1706 |
+
"learning_rate": 4.4727998242741627e-07,
|
| 1707 |
+
"loss": 3.2982,
|
| 1708 |
+
"step": 2300
|
| 1709 |
+
},
|
| 1710 |
+
{
|
| 1711 |
+
"epoch": 2.769046478198371,
|
| 1712 |
+
"grad_norm": 97.41565704345703,
|
| 1713 |
+
"learning_rate": 4.394331391017474e-07,
|
| 1714 |
+
"loss": 3.0522,
|
| 1715 |
+
"step": 2310
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 2.7810253953042645,
|
| 1719 |
+
"grad_norm": 13.397387504577637,
|
| 1720 |
+
"learning_rate": 4.316363083776766e-07,
|
| 1721 |
+
"loss": 5.0599,
|
| 1722 |
+
"step": 2320
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 2.793004312410158,
|
| 1726 |
+
"grad_norm": 14.772340774536133,
|
| 1727 |
+
"learning_rate": 4.2389018587211524e-07,
|
| 1728 |
+
"loss": 2.43,
|
| 1729 |
+
"step": 2330
|
| 1730 |
+
},
|
| 1731 |
+
{
|
| 1732 |
+
"epoch": 2.8049832295160515,
|
| 1733 |
+
"grad_norm": 218.20303344726562,
|
| 1734 |
+
"learning_rate": 4.1619546267789453e-07,
|
| 1735 |
+
"loss": 5.5137,
|
| 1736 |
+
"step": 2340
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 2.8169621466219454,
|
| 1740 |
+
"grad_norm": 312.639892578125,
|
| 1741 |
+
"learning_rate": 4.0855282530210676e-07,
|
| 1742 |
+
"loss": 4.4751,
|
| 1743 |
+
"step": 2350
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 2.828941063727839,
|
| 1747 |
+
"grad_norm": 293.28387451171875,
|
| 1748 |
+
"learning_rate": 4.0096295560485547e-07,
|
| 1749 |
+
"loss": 4.1398,
|
| 1750 |
+
"step": 2360
|
| 1751 |
+
},
|
| 1752 |
+
{
|
| 1753 |
+
"epoch": 2.840919980833733,
|
| 1754 |
+
"grad_norm": 84.95050811767578,
|
| 1755 |
+
"learning_rate": 3.934265307384239e-07,
|
| 1756 |
+
"loss": 3.7418,
|
| 1757 |
+
"step": 2370
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 2.8528988979396264,
|
| 1761 |
+
"grad_norm": 108.16950988769531,
|
| 1762 |
+
"learning_rate": 3.8594422308685793e-07,
|
| 1763 |
+
"loss": 3.392,
|
| 1764 |
+
"step": 2380
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 2.86487781504552,
|
| 1768 |
+
"grad_norm": 258.593505859375,
|
| 1769 |
+
"learning_rate": 3.785167002059799e-07,
|
| 1770 |
+
"loss": 4.247,
|
| 1771 |
+
"step": 2390
|
| 1772 |
+
},
|
| 1773 |
+
{
|
| 1774 |
+
"epoch": 2.8768567321514134,
|
| 1775 |
+
"grad_norm": 247.61813354492188,
|
| 1776 |
+
"learning_rate": 3.7114462476382966e-07,
|
| 1777 |
+
"loss": 3.9058,
|
| 1778 |
+
"step": 2400
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 2.8768567321514134,
|
| 1782 |
+
"eval_loss": 1.125927209854126,
|
| 1783 |
+
"eval_runtime": 237.8939,
|
| 1784 |
+
"eval_samples_per_second": 6.263,
|
| 1785 |
+
"eval_steps_per_second": 3.132,
|
| 1786 |
+
"step": 2400
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 2.8888356492573073,
|
| 1790 |
+
"grad_norm": 82.83909606933594,
|
| 1791 |
+
"learning_rate": 3.6382865448154187e-07,
|
| 1792 |
+
"loss": 3.9744,
|
| 1793 |
+
"step": 2410
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 2.900814566363201,
|
| 1797 |
+
"grad_norm": 289.5834655761719,
|
| 1798 |
+
"learning_rate": 3.5656944207466633e-07,
|
| 1799 |
+
"loss": 4.8423,
|
| 1800 |
+
"step": 2420
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 2.9127934834690943,
|
| 1804 |
+
"grad_norm": 11.12870979309082,
|
| 1805 |
+
"learning_rate": 3.4936763519493495e-07,
|
| 1806 |
+
"loss": 1.8868,
|
| 1807 |
+
"step": 2430
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 2.924772400574988,
|
| 1811 |
+
"grad_norm": 259.54205322265625,
|
| 1812 |
+
"learning_rate": 3.4222387637247806e-07,
|
| 1813 |
+
"loss": 5.073,
|
| 1814 |
+
"step": 2440
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 2.9367513176808817,
|
| 1818 |
+
"grad_norm": 260.9941711425781,
|
| 1819 |
+
"learning_rate": 3.351388029585007e-07,
|
| 1820 |
+
"loss": 3.9769,
|
| 1821 |
+
"step": 2450
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 2.948730234786775,
|
| 1825 |
+
"grad_norm": 138.75767517089844,
|
| 1826 |
+
"learning_rate": 3.281130470684166e-07,
|
| 1827 |
+
"loss": 1.8905,
|
| 1828 |
+
"step": 2460
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 2.9607091518926687,
|
| 1832 |
+
"grad_norm": 212.35589599609375,
|
| 1833 |
+
"learning_rate": 3.2114723552545606e-07,
|
| 1834 |
+
"loss": 5.6336,
|
| 1835 |
+
"step": 2470
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 2.9726880689985626,
|
| 1839 |
+
"grad_norm": 77.375,
|
| 1840 |
+
"learning_rate": 3.142419898047399e-07,
|
| 1841 |
+
"loss": 2.5325,
|
| 1842 |
+
"step": 2480
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 2.984666986104456,
|
| 1846 |
+
"grad_norm": 301.5721740722656,
|
| 1847 |
+
"learning_rate": 3.073979259778332e-07,
|
| 1848 |
+
"loss": 2.8451,
|
| 1849 |
+
"step": 2490
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 2.9966459032103496,
|
| 1853 |
+
"grad_norm": 23.281728744506836,
|
| 1854 |
+
"learning_rate": 3.006156546577796e-07,
|
| 1855 |
+
"loss": 4.271,
|
| 1856 |
+
"step": 2500
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 3.009583133684715,
|
| 1860 |
+
"grad_norm": 60.59037399291992,
|
| 1861 |
+
"learning_rate": 2.9389578094462607e-07,
|
| 1862 |
+
"loss": 4.768,
|
| 1863 |
+
"step": 2510
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 3.0215620507906085,
|
| 1867 |
+
"grad_norm": 12.599560737609863,
|
| 1868 |
+
"learning_rate": 2.872389043714343e-07,
|
| 1869 |
+
"loss": 3.4648,
|
| 1870 |
+
"step": 2520
|
| 1871 |
+
},
|
| 1872 |
+
{
|
| 1873 |
+
"epoch": 3.033540967896502,
|
| 1874 |
+
"grad_norm": 60.7338981628418,
|
| 1875 |
+
"learning_rate": 2.806456188507943e-07,
|
| 1876 |
+
"loss": 4.1664,
|
| 1877 |
+
"step": 2530
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 3.045519885002396,
|
| 1881 |
+
"grad_norm": 124.4323959350586,
|
| 1882 |
+
"learning_rate": 2.7411651262183465e-07,
|
| 1883 |
+
"loss": 3.2584,
|
| 1884 |
+
"step": 2540
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 3.0574988021082894,
|
| 1888 |
+
"grad_norm": 10.696157455444336,
|
| 1889 |
+
"learning_rate": 2.676521681977425e-07,
|
| 1890 |
+
"loss": 1.6209,
|
| 1891 |
+
"step": 2550
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 3.069477719214183,
|
| 1895 |
+
"grad_norm": 78.30728149414062,
|
| 1896 |
+
"learning_rate": 2.612531623137922e-07,
|
| 1897 |
+
"loss": 2.9967,
|
| 1898 |
+
"step": 2560
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 3.081456636320077,
|
| 1902 |
+
"grad_norm": 12.110499382019043,
|
| 1903 |
+
"learning_rate": 2.5492006587589033e-07,
|
| 1904 |
+
"loss": 1.8501,
|
| 1905 |
+
"step": 2570
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 3.0934355534259703,
|
| 1909 |
+
"grad_norm": 142.67459106445312,
|
| 1910 |
+
"learning_rate": 2.4865344390964016e-07,
|
| 1911 |
+
"loss": 6.7426,
|
| 1912 |
+
"step": 2580
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"epoch": 3.105414470531864,
|
| 1916 |
+
"grad_norm": 84.46195220947266,
|
| 1917 |
+
"learning_rate": 2.424538555099326e-07,
|
| 1918 |
+
"loss": 3.011,
|
| 1919 |
+
"step": 2590
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 3.1173933876377578,
|
| 1923 |
+
"grad_norm": 287.4181213378906,
|
| 1924 |
+
"learning_rate": 2.3632185379106383e-07,
|
| 1925 |
+
"loss": 4.2906,
|
| 1926 |
+
"step": 2600
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 3.1173933876377578,
|
| 1930 |
+
"eval_loss": 1.1396493911743164,
|
| 1931 |
+
"eval_runtime": 240.643,
|
| 1932 |
+
"eval_samples_per_second": 6.192,
|
| 1933 |
+
"eval_steps_per_second": 3.096,
|
| 1934 |
+
"step": 2600
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 3.1293723047436512,
|
| 1938 |
+
"grad_norm": 17.018083572387695,
|
| 1939 |
+
"learning_rate": 2.302579858373881e-07,
|
| 1940 |
+
"loss": 2.1286,
|
| 1941 |
+
"step": 2610
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 3.1413512218495447,
|
| 1945 |
+
"grad_norm": 13.265679359436035,
|
| 1946 |
+
"learning_rate": 2.2426279265450708e-07,
|
| 1947 |
+
"loss": 2.9965,
|
| 1948 |
+
"step": 2620
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"epoch": 3.1533301389554382,
|
| 1952 |
+
"grad_norm": 156.8545684814453,
|
| 1953 |
+
"learning_rate": 2.183368091210037e-07,
|
| 1954 |
+
"loss": 3.7899,
|
| 1955 |
+
"step": 2630
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 3.165309056061332,
|
| 1959 |
+
"grad_norm": 17.818618774414062,
|
| 1960 |
+
"learning_rate": 2.1248056394072078e-07,
|
| 1961 |
+
"loss": 4.1165,
|
| 1962 |
+
"step": 2640
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 3.1772879731672257,
|
| 1966 |
+
"grad_norm": 15.270909309387207,
|
| 1967 |
+
"learning_rate": 2.0669457959559177e-07,
|
| 1968 |
+
"loss": 3.1192,
|
| 1969 |
+
"step": 2650
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 3.189266890273119,
|
| 1973 |
+
"grad_norm": 78.96019744873047,
|
| 1974 |
+
"learning_rate": 2.0097937229902485e-07,
|
| 1975 |
+
"loss": 5.3403,
|
| 1976 |
+
"step": 2660
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 3.201245807379013,
|
| 1980 |
+
"grad_norm": 376.2461242675781,
|
| 1981 |
+
"learning_rate": 1.9533545194984791e-07,
|
| 1982 |
+
"loss": 3.9551,
|
| 1983 |
+
"step": 2670
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 3.2132247244849066,
|
| 1987 |
+
"grad_norm": 16.849010467529297,
|
| 1988 |
+
"learning_rate": 1.8976332208681744e-07,
|
| 1989 |
+
"loss": 5.6715,
|
| 1990 |
+
"step": 2680
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 3.2252036415908,
|
| 1994 |
+
"grad_norm": 171.6584930419922,
|
| 1995 |
+
"learning_rate": 1.8426347984369273e-07,
|
| 1996 |
+
"loss": 5.2323,
|
| 1997 |
+
"step": 2690
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 3.237182558696694,
|
| 2001 |
+
"grad_norm": 120.8387680053711,
|
| 2002 |
+
"learning_rate": 1.788364159048833e-07,
|
| 2003 |
+
"loss": 4.2853,
|
| 2004 |
+
"step": 2700
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 3.2491614758025875,
|
| 2008 |
+
"grad_norm": 13.419800758361816,
|
| 2009 |
+
"learning_rate": 1.734826144616698e-07,
|
| 2010 |
+
"loss": 2.9811,
|
| 2011 |
+
"step": 2710
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 3.261140392908481,
|
| 2015 |
+
"grad_norm": 196.2430877685547,
|
| 2016 |
+
"learning_rate": 1.6820255316900756e-07,
|
| 2017 |
+
"loss": 4.3565,
|
| 2018 |
+
"step": 2720
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 3.2731193100143745,
|
| 2022 |
+
"grad_norm": 15.63242244720459,
|
| 2023 |
+
"learning_rate": 1.6299670310290915e-07,
|
| 2024 |
+
"loss": 2.4933,
|
| 2025 |
+
"step": 2730
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 3.2850982271202684,
|
| 2029 |
+
"grad_norm": 60.21677017211914,
|
| 2030 |
+
"learning_rate": 1.5786552871841774e-07,
|
| 2031 |
+
"loss": 2.6513,
|
| 2032 |
+
"step": 2740
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 3.297077144226162,
|
| 2036 |
+
"grad_norm": 223.22097778320312,
|
| 2037 |
+
"learning_rate": 1.528094878081677e-07,
|
| 2038 |
+
"loss": 3.3477,
|
| 2039 |
+
"step": 2750
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 3.3090560613320554,
|
| 2043 |
+
"grad_norm": 95.59901428222656,
|
| 2044 |
+
"learning_rate": 1.478290314615427e-07,
|
| 2045 |
+
"loss": 3.0678,
|
| 2046 |
+
"step": 2760
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 3.3210349784379494,
|
| 2050 |
+
"grad_norm": 36.64326095581055,
|
| 2051 |
+
"learning_rate": 1.4292460402442995e-07,
|
| 2052 |
+
"loss": 3.4483,
|
| 2053 |
+
"step": 2770
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 3.333013895543843,
|
| 2057 |
+
"grad_norm": 233.20370483398438,
|
| 2058 |
+
"learning_rate": 1.3809664305957625e-07,
|
| 2059 |
+
"loss": 2.9447,
|
| 2060 |
+
"step": 2780
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 3.3449928126497364,
|
| 2064 |
+
"grad_norm": 14.824277877807617,
|
| 2065 |
+
"learning_rate": 1.3334557930754963e-07,
|
| 2066 |
+
"loss": 2.547,
|
| 2067 |
+
"step": 2790
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 3.35697172975563,
|
| 2071 |
+
"grad_norm": 208.10154724121094,
|
| 2072 |
+
"learning_rate": 1.2867183664831038e-07,
|
| 2073 |
+
"loss": 6.0572,
|
| 2074 |
+
"step": 2800
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 3.35697172975563,
|
| 2078 |
+
"eval_loss": 1.147083044052124,
|
| 2079 |
+
"eval_runtime": 239.357,
|
| 2080 |
+
"eval_samples_per_second": 6.225,
|
| 2081 |
+
"eval_steps_per_second": 3.113,
|
| 2082 |
+
"step": 2800
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 3.368950646861524,
|
| 2086 |
+
"grad_norm": 167.9786834716797,
|
| 2087 |
+
"learning_rate": 1.2407583206339256e-07,
|
| 2088 |
+
"loss": 2.6004,
|
| 2089 |
+
"step": 2810
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 3.3809295639674173,
|
| 2093 |
+
"grad_norm": 236.07484436035156,
|
| 2094 |
+
"learning_rate": 1.195579755987024e-07,
|
| 2095 |
+
"loss": 2.3534,
|
| 2096 |
+
"step": 2820
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 3.3929084810733112,
|
| 2100 |
+
"grad_norm": 38.18111801147461,
|
| 2101 |
+
"learning_rate": 1.1511867032793321e-07,
|
| 2102 |
+
"loss": 2.5498,
|
| 2103 |
+
"step": 2830
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 3.4048873981792047,
|
| 2107 |
+
"grad_norm": 136.2451934814453,
|
| 2108 |
+
"learning_rate": 1.107583123166066e-07,
|
| 2109 |
+
"loss": 5.3208,
|
| 2110 |
+
"step": 2840
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 3.416866315285098,
|
| 2114 |
+
"grad_norm": 316.2095947265625,
|
| 2115 |
+
"learning_rate": 1.0647729058673427e-07,
|
| 2116 |
+
"loss": 4.3772,
|
| 2117 |
+
"step": 2850
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 3.4288452323909917,
|
| 2121 |
+
"grad_norm": 14.228015899658203,
|
| 2122 |
+
"learning_rate": 1.0227598708211172e-07,
|
| 2123 |
+
"loss": 2.2948,
|
| 2124 |
+
"step": 2860
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 3.4408241494968856,
|
| 2128 |
+
"grad_norm": 52.19302749633789,
|
| 2129 |
+
"learning_rate": 9.81547766342401e-08,
|
| 2130 |
+
"loss": 2.223,
|
| 2131 |
+
"step": 2870
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 3.452803066602779,
|
| 2135 |
+
"grad_norm": 18.04366683959961,
|
| 2136 |
+
"learning_rate": 9.411402692888715e-08,
|
| 2137 |
+
"loss": 3.6671,
|
| 2138 |
+
"step": 2880
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 3.4647819837086726,
|
| 2142 |
+
"grad_norm": 12.00094985961914,
|
| 2143 |
+
"learning_rate": 9.015409847328037e-08,
|
| 2144 |
+
"loss": 2.3488,
|
| 2145 |
+
"step": 2890
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 3.4767609008145666,
|
| 2149 |
+
"grad_norm": 88.41423797607422,
|
| 2150 |
+
"learning_rate": 8.62753445639457e-08,
|
| 2151 |
+
"loss": 3.2758,
|
| 2152 |
+
"step": 2900
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 3.48873981792046,
|
| 2156 |
+
"grad_norm": 56.944828033447266,
|
| 2157 |
+
"learning_rate": 8.247811125518489e-08,
|
| 2158 |
+
"loss": 1.576,
|
| 2159 |
+
"step": 2910
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 3.5007187350263536,
|
| 2163 |
+
"grad_norm": 323.6489562988281,
|
| 2164 |
+
"learning_rate": 7.876273732820327e-08,
|
| 2165 |
+
"loss": 4.9528,
|
| 2166 |
+
"step": 2920
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 3.512697652132247,
|
| 2170 |
+
"grad_norm": 10.932809829711914,
|
| 2171 |
+
"learning_rate": 7.51295542608834e-08,
|
| 2172 |
+
"loss": 3.6918,
|
| 2173 |
+
"step": 2930
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 3.524676569238141,
|
| 2177 |
+
"grad_norm": 12.945392608642578,
|
| 2178 |
+
"learning_rate": 7.157888619821106e-08,
|
| 2179 |
+
"loss": 1.7161,
|
| 2180 |
+
"step": 2940
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 3.5366554863440345,
|
| 2184 |
+
"grad_norm": 358.42498779296875,
|
| 2185 |
+
"learning_rate": 6.811104992335648e-08,
|
| 2186 |
+
"loss": 4.4565,
|
| 2187 |
+
"step": 2950
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 3.548634403449928,
|
| 2191 |
+
"grad_norm": 231.78411865234375,
|
| 2192 |
+
"learning_rate": 6.47263548294108e-08,
|
| 2193 |
+
"loss": 3.5586,
|
| 2194 |
+
"step": 2960
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 3.560613320555822,
|
| 2198 |
+
"grad_norm": 231.84341430664062,
|
| 2199 |
+
"learning_rate": 6.142510289178337e-08,
|
| 2200 |
+
"loss": 3.4724,
|
| 2201 |
+
"step": 2970
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 3.5725922376617154,
|
| 2205 |
+
"grad_norm": 198.78810119628906,
|
| 2206 |
+
"learning_rate": 5.8207588641260185e-08,
|
| 2207 |
+
"loss": 2.5415,
|
| 2208 |
+
"step": 2980
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 3.584571154767609,
|
| 2212 |
+
"grad_norm": 76.49454498291016,
|
| 2213 |
+
"learning_rate": 5.507409913772543e-08,
|
| 2214 |
+
"loss": 3.7494,
|
| 2215 |
+
"step": 2990
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 3.5965500718735024,
|
| 2219 |
+
"grad_norm": 262.359130859375,
|
| 2220 |
+
"learning_rate": 5.202491394455155e-08,
|
| 2221 |
+
"loss": 4.0544,
|
| 2222 |
+
"step": 3000
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 3.5965500718735024,
|
| 2226 |
+
"eval_loss": 1.1526151895523071,
|
| 2227 |
+
"eval_runtime": 238.4822,
|
| 2228 |
+
"eval_samples_per_second": 6.248,
|
| 2229 |
+
"eval_steps_per_second": 3.124,
|
| 2230 |
+
"step": 3000
|
| 2231 |
+
},
|
| 2232 |
+
{
|
| 2233 |
+
"epoch": 3.6085289889793963,
|
| 2234 |
+
"grad_norm": 14.773038864135742,
|
| 2235 |
+
"learning_rate": 4.9060305103657e-08,
|
| 2236 |
+
"loss": 3.8126,
|
| 2237 |
+
"step": 3010
|
| 2238 |
+
},
|
| 2239 |
+
{
|
| 2240 |
+
"epoch": 3.62050790608529,
|
| 2241 |
+
"grad_norm": 288.3175354003906,
|
| 2242 |
+
"learning_rate": 4.61805371112356e-08,
|
| 2243 |
+
"loss": 2.2371,
|
| 2244 |
+
"step": 3020
|
| 2245 |
+
},
|
| 2246 |
+
{
|
| 2247 |
+
"epoch": 3.6324868231911838,
|
| 2248 |
+
"grad_norm": 15.71839427947998,
|
| 2249 |
+
"learning_rate": 4.3661497350331423e-08,
|
| 2250 |
+
"loss": 2.8255,
|
| 2251 |
+
"step": 3030
|
| 2252 |
+
},
|
| 2253 |
+
{
|
| 2254 |
+
"epoch": 3.6444657402970773,
|
| 2255 |
+
"grad_norm": 47.34138870239258,
|
| 2256 |
+
"learning_rate": 4.094362852900846e-08,
|
| 2257 |
+
"loss": 2.4564,
|
| 2258 |
+
"step": 3040
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 3.6564446574029708,
|
| 2262 |
+
"grad_norm": 133.03443908691406,
|
| 2263 |
+
"learning_rate": 3.8311324709047524e-08,
|
| 2264 |
+
"loss": 3.9076,
|
| 2265 |
+
"step": 3050
|
| 2266 |
+
},
|
| 2267 |
+
{
|
| 2268 |
+
"epoch": 3.6684235745088642,
|
| 2269 |
+
"grad_norm": 85.79124450683594,
|
| 2270 |
+
"learning_rate": 3.57648207390836e-08,
|
| 2271 |
+
"loss": 1.6264,
|
| 2272 |
+
"step": 3060
|
| 2273 |
+
},
|
| 2274 |
+
{
|
| 2275 |
+
"epoch": 3.680402491614758,
|
| 2276 |
+
"grad_norm": 11.73284912109375,
|
| 2277 |
+
"learning_rate": 3.3304343812869175e-08,
|
| 2278 |
+
"loss": 2.2377,
|
| 2279 |
+
"step": 3070
|
| 2280 |
+
},
|
| 2281 |
+
{
|
| 2282 |
+
"epoch": 3.6923814087206517,
|
| 2283 |
+
"grad_norm": 103.52379608154297,
|
| 2284 |
+
"learning_rate": 3.0930113449003536e-08,
|
| 2285 |
+
"loss": 2.5226,
|
| 2286 |
+
"step": 3080
|
| 2287 |
+
},
|
| 2288 |
+
{
|
| 2289 |
+
"epoch": 3.704360325826545,
|
| 2290 |
+
"grad_norm": 121.46673583984375,
|
| 2291 |
+
"learning_rate": 2.8642341471348585e-08,
|
| 2292 |
+
"loss": 5.8129,
|
| 2293 |
+
"step": 3090
|
| 2294 |
+
},
|
| 2295 |
+
{
|
| 2296 |
+
"epoch": 3.716339242932439,
|
| 2297 |
+
"grad_norm": 39.363887786865234,
|
| 2298 |
+
"learning_rate": 2.644123199013004e-08,
|
| 2299 |
+
"loss": 1.8684,
|
| 2300 |
+
"step": 3100
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 3.7283181600383326,
|
| 2304 |
+
"grad_norm": 22.17119598388672,
|
| 2305 |
+
"learning_rate": 2.432698138372713e-08,
|
| 2306 |
+
"loss": 4.5753,
|
| 2307 |
+
"step": 3110
|
| 2308 |
+
},
|
| 2309 |
+
{
|
| 2310 |
+
"epoch": 3.740297077144226,
|
| 2311 |
+
"grad_norm": 123.43873596191406,
|
| 2312 |
+
"learning_rate": 2.2299778281151927e-08,
|
| 2313 |
+
"loss": 3.4706,
|
| 2314 |
+
"step": 3120
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"epoch": 3.7522759942501196,
|
| 2318 |
+
"grad_norm": 62.01347351074219,
|
| 2319 |
+
"learning_rate": 2.03598035452206e-08,
|
| 2320 |
+
"loss": 1.0962,
|
| 2321 |
+
"step": 3130
|
| 2322 |
+
},
|
| 2323 |
+
{
|
| 2324 |
+
"epoch": 3.7642549113560135,
|
| 2325 |
+
"grad_norm": 106.05194091796875,
|
| 2326 |
+
"learning_rate": 1.8507230256417316e-08,
|
| 2327 |
+
"loss": 4.0847,
|
| 2328 |
+
"step": 3140
|
| 2329 |
+
},
|
| 2330 |
+
{
|
| 2331 |
+
"epoch": 3.776233828461907,
|
| 2332 |
+
"grad_norm": 14.505044937133789,
|
| 2333 |
+
"learning_rate": 1.674222369745182e-08,
|
| 2334 |
+
"loss": 2.7005,
|
| 2335 |
+
"step": 3150
|
| 2336 |
+
},
|
| 2337 |
+
{
|
| 2338 |
+
"epoch": 3.7882127455678005,
|
| 2339 |
+
"grad_norm": 74.14771270751953,
|
| 2340 |
+
"learning_rate": 1.5064941338513548e-08,
|
| 2341 |
+
"loss": 4.5833,
|
| 2342 |
+
"step": 3160
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 3.8001916626736945,
|
| 2346 |
+
"grad_norm": 109.61576080322266,
|
| 2347 |
+
"learning_rate": 1.3475532823222779e-08,
|
| 2348 |
+
"loss": 3.3511,
|
| 2349 |
+
"step": 3170
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 3.812170579779588,
|
| 2353 |
+
"grad_norm": 142.658447265625,
|
| 2354 |
+
"learning_rate": 1.1974139955279294e-08,
|
| 2355 |
+
"loss": 4.0569,
|
| 2356 |
+
"step": 3180
|
| 2357 |
+
},
|
| 2358 |
+
{
|
| 2359 |
+
"epoch": 3.8241494968854814,
|
| 2360 |
+
"grad_norm": 374.00701904296875,
|
| 2361 |
+
"learning_rate": 1.0560896685811061e-08,
|
| 2362 |
+
"loss": 3.657,
|
| 2363 |
+
"step": 3190
|
| 2364 |
+
},
|
| 2365 |
+
{
|
| 2366 |
+
"epoch": 3.836128413991375,
|
| 2367 |
+
"grad_norm": 191.76698303222656,
|
| 2368 |
+
"learning_rate": 9.235929101423457e-09,
|
| 2369 |
+
"loss": 3.2204,
|
| 2370 |
+
"step": 3200
|
| 2371 |
+
},
|
| 2372 |
+
{
|
| 2373 |
+
"epoch": 3.836128413991375,
|
| 2374 |
+
"eval_loss": 1.1512279510498047,
|
| 2375 |
+
"eval_runtime": 239.5597,
|
| 2376 |
+
"eval_samples_per_second": 6.22,
|
| 2377 |
+
"eval_steps_per_second": 3.11,
|
| 2378 |
+
"step": 3200
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 3.848107331097269,
|
| 2382 |
+
"grad_norm": 89.6385726928711,
|
| 2383 |
+
"learning_rate": 7.99935541295016e-09,
|
| 2384 |
+
"loss": 4.2172,
|
| 2385 |
+
"step": 3210
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 3.8600862482031624,
|
| 2389 |
+
"grad_norm": 52.579227447509766,
|
| 2390 |
+
"learning_rate": 6.8512859449064705e-09,
|
| 2391 |
+
"loss": 3.1792,
|
| 2392 |
+
"step": 3220
|
| 2393 |
+
},
|
| 2394 |
+
{
|
| 2395 |
+
"epoch": 3.8720651653090563,
|
| 2396 |
+
"grad_norm": 159.96087646484375,
|
| 2397 |
+
"learning_rate": 5.791823125646522e-09,
|
| 2398 |
+
"loss": 4.6078,
|
| 2399 |
+
"step": 3230
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 3.88404408241495,
|
| 2403 |
+
"grad_norm": 248.06581115722656,
|
| 2404 |
+
"learning_rate": 4.8210614782245866e-09,
|
| 2405 |
+
"loss": 4.6227,
|
| 2406 |
+
"step": 3240
|
| 2407 |
+
},
|
| 2408 |
+
{
|
| 2409 |
+
"epoch": 3.8960229995208433,
|
| 2410 |
+
"grad_norm": 10.945176124572754,
|
| 2411 |
+
"learning_rate": 3.939087611962377e-09,
|
| 2412 |
+
"loss": 1.8884,
|
| 2413 |
+
"step": 3250
|
| 2414 |
+
},
|
| 2415 |
+
{
|
| 2416 |
+
"epoch": 3.908001916626737,
|
| 2417 |
+
"grad_norm": 279.2950439453125,
|
| 2418 |
+
"learning_rate": 3.1459802147214554e-09,
|
| 2419 |
+
"loss": 3.5358,
|
| 2420 |
+
"step": 3260
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 3.9199808337326307,
|
| 2424 |
+
"grad_norm": 278.5760192871094,
|
| 2425 |
+
"learning_rate": 2.441810045883175e-09,
|
| 2426 |
+
"loss": 4.0377,
|
| 2427 |
+
"step": 3270
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 3.9319597508385242,
|
| 2431 |
+
"grad_norm": 20.70584487915039,
|
| 2432 |
+
"learning_rate": 1.8266399300355118e-09,
|
| 2433 |
+
"loss": 4.9587,
|
| 2434 |
+
"step": 3280
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"epoch": 3.9439386679444177,
|
| 2438 |
+
"grad_norm": 238.9534912109375,
|
| 2439 |
+
"learning_rate": 1.300524751368326e-09,
|
| 2440 |
+
"loss": 3.8602,
|
| 2441 |
+
"step": 3290
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 3.9559175850503117,
|
| 2445 |
+
"grad_norm": 16.01015281677246,
|
| 2446 |
+
"learning_rate": 8.635114487760553e-10,
|
| 2447 |
+
"loss": 1.4575,
|
| 2448 |
+
"step": 3300
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 3.967896502156205,
|
| 2452 |
+
"grad_norm": 14.110264778137207,
|
| 2453 |
+
"learning_rate": 5.156390116707321e-10,
|
| 2454 |
+
"loss": 1.1673,
|
| 2455 |
+
"step": 3310
|
| 2456 |
+
},
|
| 2457 |
+
{
|
| 2458 |
+
"epoch": 3.9798754192620986,
|
| 2459 |
+
"grad_norm": 43.93592834472656,
|
| 2460 |
+
"learning_rate": 2.56938476502655e-10,
|
| 2461 |
+
"loss": 3.8769,
|
| 2462 |
+
"step": 3320
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 3.991854336367992,
|
| 2466 |
+
"grad_norm": 9.19588851928711,
|
| 2467 |
+
"learning_rate": 8.743292399204793e-11,
|
| 2468 |
+
"loss": 2.4475,
|
| 2469 |
+
"step": 3330
|
| 2470 |
+
}
|
| 2471 |
+
],
|
| 2472 |
+
"logging_steps": 10,
|
| 2473 |
+
"max_steps": 3336,
|
| 2474 |
+
"num_input_tokens_seen": 0,
|
| 2475 |
+
"num_train_epochs": 4,
|
| 2476 |
+
"save_steps": 300,
|
| 2477 |
+
"stateful_callbacks": {
|
| 2478 |
+
"TrainerControl": {
|
| 2479 |
+
"args": {
|
| 2480 |
+
"should_epoch_stop": false,
|
| 2481 |
+
"should_evaluate": false,
|
| 2482 |
+
"should_log": false,
|
| 2483 |
+
"should_save": true,
|
| 2484 |
+
"should_training_stop": true
|
| 2485 |
+
},
|
| 2486 |
+
"attributes": {}
|
| 2487 |
+
}
|
| 2488 |
+
},
|
| 2489 |
+
"total_flos": 7.982695446856335e+17,
|
| 2490 |
+
"train_batch_size": 2,
|
| 2491 |
+
"trial_name": null,
|
| 2492 |
+
"trial_params": null
|
| 2493 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a8191a575d43cd30b60263c09bf397b3555e3aafcb9b955a356e0891484ee671
|
| 3 |
+
size 5344
|