Model save
Browse files- README.md +73 -0
- adapter_config.json +25 -0
- adapter_model.safetensors +3 -0
- all_results.json +21 -0
- eval_results.json +16 -0
- special_tokens_map.json +30 -0
- tokenizer.json +0 -0
- tokenizer_config.json +41 -0
- train_results.json +8 -0
- trainer_state.json +1098 -0
- training_args.bin +3 -0
README.md
ADDED
|
@@ -0,0 +1,73 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
base_model: mistralai/Mistral-7B-v0.1
|
| 4 |
+
tags:
|
| 5 |
+
- generated_from_trainer
|
| 6 |
+
model-index:
|
| 7 |
+
- name: zephyr-7b-dpo-lora
|
| 8 |
+
results: []
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 12 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 13 |
+
|
| 14 |
+
# zephyr-7b-dpo-lora
|
| 15 |
+
|
| 16 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on the None dataset.
|
| 17 |
+
It achieves the following results on the evaluation set:
|
| 18 |
+
- Loss: 0.5657
|
| 19 |
+
- Rewards/chosen: -0.0834
|
| 20 |
+
- Rewards/rejected: -0.5329
|
| 21 |
+
- Rewards/accuracies: 0.7020
|
| 22 |
+
- Rewards/margins: 0.4496
|
| 23 |
+
- Logps/rejected: -224.6124
|
| 24 |
+
- Logps/chosen: -265.4659
|
| 25 |
+
- Logits/rejected: -2.0001
|
| 26 |
+
- Logits/chosen: -2.1212
|
| 27 |
+
|
| 28 |
+
## Model description
|
| 29 |
+
|
| 30 |
+
More information needed
|
| 31 |
+
|
| 32 |
+
## Intended uses & limitations
|
| 33 |
+
|
| 34 |
+
More information needed
|
| 35 |
+
|
| 36 |
+
## Training and evaluation data
|
| 37 |
+
|
| 38 |
+
More information needed
|
| 39 |
+
|
| 40 |
+
## Training procedure
|
| 41 |
+
|
| 42 |
+
### Training hyperparameters
|
| 43 |
+
|
| 44 |
+
The following hyperparameters were used during training:
|
| 45 |
+
- learning_rate: 5e-07
|
| 46 |
+
- train_batch_size: 2
|
| 47 |
+
- eval_batch_size: 4
|
| 48 |
+
- seed: 42
|
| 49 |
+
- distributed_type: multi-GPU
|
| 50 |
+
- num_devices: 4
|
| 51 |
+
- gradient_accumulation_steps: 32
|
| 52 |
+
- total_train_batch_size: 256
|
| 53 |
+
- total_eval_batch_size: 16
|
| 54 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 55 |
+
- lr_scheduler_type: linear
|
| 56 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 57 |
+
- num_epochs: 3
|
| 58 |
+
|
| 59 |
+
### Training results
|
| 60 |
+
|
| 61 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 62 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 63 |
+
| 0.6123 | 1.0 | 242 | 0.6065 | 0.0003 | -0.2418 | 0.6660 | 0.2421 | -221.7009 | -264.6292 | -2.0235 | -2.1436 |
|
| 64 |
+
| 0.5841 | 2.0 | 484 | 0.5733 | -0.0606 | -0.4698 | 0.7040 | 0.4092 | -223.9813 | -265.2384 | -2.0064 | -2.1272 |
|
| 65 |
+
| 0.567 | 3.0 | 726 | 0.5657 | -0.0834 | -0.5329 | 0.7020 | 0.4496 | -224.6124 | -265.4659 | -2.0001 | -2.1212 |
|
| 66 |
+
|
| 67 |
+
|
| 68 |
+
### Framework versions
|
| 69 |
+
|
| 70 |
+
- Transformers 4.35.0
|
| 71 |
+
- Pytorch 2.0.1
|
| 72 |
+
- Datasets 2.14.6
|
| 73 |
+
- Tokenizers 0.14.1
|
adapter_config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "mistralai/Mistral-7B-v0.1",
|
| 5 |
+
"bias": "none",
|
| 6 |
+
"fan_in_fan_out": false,
|
| 7 |
+
"inference_mode": true,
|
| 8 |
+
"init_lora_weights": true,
|
| 9 |
+
"layers_pattern": null,
|
| 10 |
+
"layers_to_transform": null,
|
| 11 |
+
"lora_alpha": 16,
|
| 12 |
+
"lora_dropout": 0.1,
|
| 13 |
+
"modules_to_save": null,
|
| 14 |
+
"peft_type": "LORA",
|
| 15 |
+
"r": 64,
|
| 16 |
+
"rank_pattern": {},
|
| 17 |
+
"revision": null,
|
| 18 |
+
"target_modules": [
|
| 19 |
+
"v_proj",
|
| 20 |
+
"o_proj",
|
| 21 |
+
"k_proj",
|
| 22 |
+
"q_proj"
|
| 23 |
+
],
|
| 24 |
+
"task_type": "CAUSAL_LM"
|
| 25 |
+
}
|
adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3e7a5241ce92e10bd13a1420f6b4bcf5bc57f917f958a8d5fedf2c6ac50f9e09
|
| 3 |
+
size 218138576
|
all_results.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_logits/chosen": -2.1211588382720947,
|
| 4 |
+
"eval_logits/rejected": -2.000145435333252,
|
| 5 |
+
"eval_logps/chosen": -265.46588134765625,
|
| 6 |
+
"eval_logps/rejected": -224.6123809814453,
|
| 7 |
+
"eval_loss": 0.5657259225845337,
|
| 8 |
+
"eval_rewards/accuracies": 0.7020000219345093,
|
| 9 |
+
"eval_rewards/chosen": -0.08337792754173279,
|
| 10 |
+
"eval_rewards/margins": 0.44957080483436584,
|
| 11 |
+
"eval_rewards/rejected": -0.5329487323760986,
|
| 12 |
+
"eval_runtime": 600.8868,
|
| 13 |
+
"eval_samples": 2000,
|
| 14 |
+
"eval_samples_per_second": 3.328,
|
| 15 |
+
"eval_steps_per_second": 0.208,
|
| 16 |
+
"train_loss": 0.6034470564241908,
|
| 17 |
+
"train_runtime": 84521.5132,
|
| 18 |
+
"train_samples": 61966,
|
| 19 |
+
"train_samples_per_second": 2.199,
|
| 20 |
+
"train_steps_per_second": 0.009
|
| 21 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_logits/chosen": -2.1211588382720947,
|
| 4 |
+
"eval_logits/rejected": -2.000145435333252,
|
| 5 |
+
"eval_logps/chosen": -265.46588134765625,
|
| 6 |
+
"eval_logps/rejected": -224.6123809814453,
|
| 7 |
+
"eval_loss": 0.5657259225845337,
|
| 8 |
+
"eval_rewards/accuracies": 0.7020000219345093,
|
| 9 |
+
"eval_rewards/chosen": -0.08337792754173279,
|
| 10 |
+
"eval_rewards/margins": 0.44957080483436584,
|
| 11 |
+
"eval_rewards/rejected": -0.5329487323760986,
|
| 12 |
+
"eval_runtime": 600.8868,
|
| 13 |
+
"eval_samples": 2000,
|
| 14 |
+
"eval_samples_per_second": 3.328,
|
| 15 |
+
"eval_steps_per_second": 0.208
|
| 16 |
+
}
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "<s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "</s>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": {
|
| 17 |
+
"content": "</s>",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"unk_token": {
|
| 24 |
+
"content": "<unk>",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
}
|
| 30 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "<unk>",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "<s>",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "</s>",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
}
|
| 27 |
+
},
|
| 28 |
+
"additional_special_tokens": [],
|
| 29 |
+
"bos_token": "<s>",
|
| 30 |
+
"chat_template": "{% for message in messages %}\n{% if message['role'] == 'user' %}\n{{ '<|user|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'system' %}\n{{ '<|system|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'assistant' %}\n{{ '<|assistant|>\n' + message['content'] + eos_token }}\n{% endif %}\n{% if loop.last and add_generation_prompt %}\n{{ '<|assistant|>' }}\n{% endif %}\n{% endfor %}",
|
| 31 |
+
"clean_up_tokenization_spaces": false,
|
| 32 |
+
"eos_token": "</s>",
|
| 33 |
+
"legacy": true,
|
| 34 |
+
"model_max_length": 2048,
|
| 35 |
+
"pad_token": "</s>",
|
| 36 |
+
"sp_model_kwargs": {},
|
| 37 |
+
"spaces_between_special_tokens": false,
|
| 38 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 39 |
+
"unk_token": "<unk>",
|
| 40 |
+
"use_default_system_prompt": true
|
| 41 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"train_loss": 0.6034470564241908,
|
| 4 |
+
"train_runtime": 84521.5132,
|
| 5 |
+
"train_samples": 61966,
|
| 6 |
+
"train_samples_per_second": 2.199,
|
| 7 |
+
"train_steps_per_second": 0.009
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1098 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.9992254066615027,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 726,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 6.84931506849315e-09,
|
| 14 |
+
"logits/chosen": -2.34800124168396,
|
| 15 |
+
"logits/rejected": -2.4178409576416016,
|
| 16 |
+
"logps/chosen": -271.47698974609375,
|
| 17 |
+
"logps/rejected": -208.94898986816406,
|
| 18 |
+
"loss": 0.6931,
|
| 19 |
+
"rewards/accuracies": 0.0,
|
| 20 |
+
"rewards/chosen": 0.0,
|
| 21 |
+
"rewards/margins": 0.0,
|
| 22 |
+
"rewards/rejected": 0.0,
|
| 23 |
+
"step": 1
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.04,
|
| 27 |
+
"learning_rate": 6.84931506849315e-08,
|
| 28 |
+
"logits/chosen": -2.4227354526519775,
|
| 29 |
+
"logits/rejected": -2.355938196182251,
|
| 30 |
+
"logps/chosen": -293.4179992675781,
|
| 31 |
+
"logps/rejected": -226.30238342285156,
|
| 32 |
+
"loss": 0.6928,
|
| 33 |
+
"rewards/accuracies": 0.4496527910232544,
|
| 34 |
+
"rewards/chosen": 0.0005499552935361862,
|
| 35 |
+
"rewards/margins": 0.00035988984745927155,
|
| 36 |
+
"rewards/rejected": 0.00019006534421350807,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.08,
|
| 41 |
+
"learning_rate": 1.36986301369863e-07,
|
| 42 |
+
"logits/chosen": -2.431461811065674,
|
| 43 |
+
"logits/rejected": -2.4046578407287598,
|
| 44 |
+
"logps/chosen": -278.50103759765625,
|
| 45 |
+
"logps/rejected": -216.76806640625,
|
| 46 |
+
"loss": 0.6931,
|
| 47 |
+
"rewards/accuracies": 0.518750011920929,
|
| 48 |
+
"rewards/chosen": 0.004533737897872925,
|
| 49 |
+
"rewards/margins": 0.002973187016323209,
|
| 50 |
+
"rewards/rejected": 0.0015605507651343942,
|
| 51 |
+
"step": 20
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.12,
|
| 55 |
+
"learning_rate": 2.054794520547945e-07,
|
| 56 |
+
"logits/chosen": -2.389216899871826,
|
| 57 |
+
"logits/rejected": -2.3487401008605957,
|
| 58 |
+
"logps/chosen": -252.997314453125,
|
| 59 |
+
"logps/rejected": -207.15963745117188,
|
| 60 |
+
"loss": 0.692,
|
| 61 |
+
"rewards/accuracies": 0.53125,
|
| 62 |
+
"rewards/chosen": 0.0029909531585872173,
|
| 63 |
+
"rewards/margins": 0.00379578466527164,
|
| 64 |
+
"rewards/rejected": -0.000804831855930388,
|
| 65 |
+
"step": 30
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.17,
|
| 69 |
+
"learning_rate": 2.73972602739726e-07,
|
| 70 |
+
"logits/chosen": -2.4600839614868164,
|
| 71 |
+
"logits/rejected": -2.4145703315734863,
|
| 72 |
+
"logps/chosen": -283.77569580078125,
|
| 73 |
+
"logps/rejected": -216.48251342773438,
|
| 74 |
+
"loss": 0.6909,
|
| 75 |
+
"rewards/accuracies": 0.520312488079071,
|
| 76 |
+
"rewards/chosen": 0.0019030813127756119,
|
| 77 |
+
"rewards/margins": 0.005452433601021767,
|
| 78 |
+
"rewards/rejected": -0.0035493518225848675,
|
| 79 |
+
"step": 40
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.21,
|
| 83 |
+
"learning_rate": 3.424657534246575e-07,
|
| 84 |
+
"logits/chosen": -2.436392307281494,
|
| 85 |
+
"logits/rejected": -2.3965039253234863,
|
| 86 |
+
"logps/chosen": -267.26617431640625,
|
| 87 |
+
"logps/rejected": -223.65829467773438,
|
| 88 |
+
"loss": 0.6898,
|
| 89 |
+
"rewards/accuracies": 0.5703125,
|
| 90 |
+
"rewards/chosen": 0.0054888492450118065,
|
| 91 |
+
"rewards/margins": 0.009707379154860973,
|
| 92 |
+
"rewards/rejected": -0.004218529909849167,
|
| 93 |
+
"step": 50
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.25,
|
| 97 |
+
"learning_rate": 4.10958904109589e-07,
|
| 98 |
+
"logits/chosen": -2.4120891094207764,
|
| 99 |
+
"logits/rejected": -2.388906478881836,
|
| 100 |
+
"logps/chosen": -266.8415832519531,
|
| 101 |
+
"logps/rejected": -214.7452850341797,
|
| 102 |
+
"loss": 0.6867,
|
| 103 |
+
"rewards/accuracies": 0.581250011920929,
|
| 104 |
+
"rewards/chosen": 0.006448288913816214,
|
| 105 |
+
"rewards/margins": 0.012911155819892883,
|
| 106 |
+
"rewards/rejected": -0.006462865974754095,
|
| 107 |
+
"step": 60
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.29,
|
| 111 |
+
"learning_rate": 4.794520547945205e-07,
|
| 112 |
+
"logits/chosen": -2.3903424739837646,
|
| 113 |
+
"logits/rejected": -2.397461414337158,
|
| 114 |
+
"logps/chosen": -254.0715789794922,
|
| 115 |
+
"logps/rejected": -214.6645050048828,
|
| 116 |
+
"loss": 0.6826,
|
| 117 |
+
"rewards/accuracies": 0.5843750238418579,
|
| 118 |
+
"rewards/chosen": 0.00976890604943037,
|
| 119 |
+
"rewards/margins": 0.019091714173555374,
|
| 120 |
+
"rewards/rejected": -0.00932280719280243,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.33,
|
| 125 |
+
"learning_rate": 4.946401225114854e-07,
|
| 126 |
+
"logits/chosen": -2.4295127391815186,
|
| 127 |
+
"logits/rejected": -2.3777496814727783,
|
| 128 |
+
"logps/chosen": -265.0728759765625,
|
| 129 |
+
"logps/rejected": -218.78775024414062,
|
| 130 |
+
"loss": 0.6796,
|
| 131 |
+
"rewards/accuracies": 0.6328125,
|
| 132 |
+
"rewards/chosen": 0.01282783318310976,
|
| 133 |
+
"rewards/margins": 0.030523013323545456,
|
| 134 |
+
"rewards/rejected": -0.017695177346467972,
|
| 135 |
+
"step": 80
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.37,
|
| 139 |
+
"learning_rate": 4.869831546707504e-07,
|
| 140 |
+
"logits/chosen": -2.4791455268859863,
|
| 141 |
+
"logits/rejected": -2.4226763248443604,
|
| 142 |
+
"logps/chosen": -271.70550537109375,
|
| 143 |
+
"logps/rejected": -224.4741973876953,
|
| 144 |
+
"loss": 0.6724,
|
| 145 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 146 |
+
"rewards/chosen": 0.025005927309393883,
|
| 147 |
+
"rewards/margins": 0.04857773706316948,
|
| 148 |
+
"rewards/rejected": -0.023571809753775597,
|
| 149 |
+
"step": 90
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.41,
|
| 153 |
+
"learning_rate": 4.793261868300153e-07,
|
| 154 |
+
"logits/chosen": -2.417412042617798,
|
| 155 |
+
"logits/rejected": -2.403857707977295,
|
| 156 |
+
"logps/chosen": -273.85968017578125,
|
| 157 |
+
"logps/rejected": -227.78335571289062,
|
| 158 |
+
"loss": 0.6681,
|
| 159 |
+
"rewards/accuracies": 0.6390625238418579,
|
| 160 |
+
"rewards/chosen": 0.028032511472702026,
|
| 161 |
+
"rewards/margins": 0.058635223656892776,
|
| 162 |
+
"rewards/rejected": -0.030602704733610153,
|
| 163 |
+
"step": 100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.45,
|
| 167 |
+
"learning_rate": 4.7166921898928023e-07,
|
| 168 |
+
"logits/chosen": -2.4506094455718994,
|
| 169 |
+
"logits/rejected": -2.390385866165161,
|
| 170 |
+
"logps/chosen": -252.6039276123047,
|
| 171 |
+
"logps/rejected": -222.64761352539062,
|
| 172 |
+
"loss": 0.6624,
|
| 173 |
+
"rewards/accuracies": 0.6578124761581421,
|
| 174 |
+
"rewards/chosen": 0.024289341643452644,
|
| 175 |
+
"rewards/margins": 0.06630216538906097,
|
| 176 |
+
"rewards/rejected": -0.04201282188296318,
|
| 177 |
+
"step": 110
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.5,
|
| 181 |
+
"learning_rate": 4.640122511485451e-07,
|
| 182 |
+
"logits/chosen": -2.4121220111846924,
|
| 183 |
+
"logits/rejected": -2.4016165733337402,
|
| 184 |
+
"logps/chosen": -256.88958740234375,
|
| 185 |
+
"logps/rejected": -219.8152618408203,
|
| 186 |
+
"loss": 0.6584,
|
| 187 |
+
"rewards/accuracies": 0.6265624761581421,
|
| 188 |
+
"rewards/chosen": 0.031187813729047775,
|
| 189 |
+
"rewards/margins": 0.08014924824237823,
|
| 190 |
+
"rewards/rejected": -0.04896143823862076,
|
| 191 |
+
"step": 120
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.54,
|
| 195 |
+
"learning_rate": 4.563552833078101e-07,
|
| 196 |
+
"logits/chosen": -2.456305980682373,
|
| 197 |
+
"logits/rejected": -2.394624948501587,
|
| 198 |
+
"logps/chosen": -265.97003173828125,
|
| 199 |
+
"logps/rejected": -225.9492950439453,
|
| 200 |
+
"loss": 0.6492,
|
| 201 |
+
"rewards/accuracies": 0.682812511920929,
|
| 202 |
+
"rewards/chosen": 0.02872040495276451,
|
| 203 |
+
"rewards/margins": 0.09756486117839813,
|
| 204 |
+
"rewards/rejected": -0.06884445250034332,
|
| 205 |
+
"step": 130
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.58,
|
| 209 |
+
"learning_rate": 4.4869831546707505e-07,
|
| 210 |
+
"logits/chosen": -2.460322380065918,
|
| 211 |
+
"logits/rejected": -2.4006354808807373,
|
| 212 |
+
"logps/chosen": -270.3127746582031,
|
| 213 |
+
"logps/rejected": -230.38339233398438,
|
| 214 |
+
"loss": 0.6493,
|
| 215 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 216 |
+
"rewards/chosen": 0.03178644925355911,
|
| 217 |
+
"rewards/margins": 0.11458346992731094,
|
| 218 |
+
"rewards/rejected": -0.08279702812433243,
|
| 219 |
+
"step": 140
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.62,
|
| 223 |
+
"learning_rate": 4.4104134762633994e-07,
|
| 224 |
+
"logits/chosen": -2.484839916229248,
|
| 225 |
+
"logits/rejected": -2.4318273067474365,
|
| 226 |
+
"logps/chosen": -265.9433288574219,
|
| 227 |
+
"logps/rejected": -226.5823974609375,
|
| 228 |
+
"loss": 0.6432,
|
| 229 |
+
"rewards/accuracies": 0.703125,
|
| 230 |
+
"rewards/chosen": 0.03730004280805588,
|
| 231 |
+
"rewards/margins": 0.12808530032634735,
|
| 232 |
+
"rewards/rejected": -0.09078525006771088,
|
| 233 |
+
"step": 150
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.66,
|
| 237 |
+
"learning_rate": 4.333843797856049e-07,
|
| 238 |
+
"logits/chosen": -2.44659161567688,
|
| 239 |
+
"logits/rejected": -2.402761936187744,
|
| 240 |
+
"logps/chosen": -279.3069152832031,
|
| 241 |
+
"logps/rejected": -239.0000457763672,
|
| 242 |
+
"loss": 0.6334,
|
| 243 |
+
"rewards/accuracies": 0.6890624761581421,
|
| 244 |
+
"rewards/chosen": 0.04617582634091377,
|
| 245 |
+
"rewards/margins": 0.15378056466579437,
|
| 246 |
+
"rewards/rejected": -0.1076047420501709,
|
| 247 |
+
"step": 160
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.7,
|
| 251 |
+
"learning_rate": 4.257274119448698e-07,
|
| 252 |
+
"logits/chosen": -2.47914457321167,
|
| 253 |
+
"logits/rejected": -2.4519801139831543,
|
| 254 |
+
"logps/chosen": -270.73345947265625,
|
| 255 |
+
"logps/rejected": -239.12258911132812,
|
| 256 |
+
"loss": 0.6333,
|
| 257 |
+
"rewards/accuracies": 0.660937488079071,
|
| 258 |
+
"rewards/chosen": 0.02273646369576454,
|
| 259 |
+
"rewards/margins": 0.1370951235294342,
|
| 260 |
+
"rewards/rejected": -0.11435866355895996,
|
| 261 |
+
"step": 170
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.74,
|
| 265 |
+
"learning_rate": 4.180704441041347e-07,
|
| 266 |
+
"logits/chosen": -2.435364246368408,
|
| 267 |
+
"logits/rejected": -2.388087272644043,
|
| 268 |
+
"logps/chosen": -256.7988586425781,
|
| 269 |
+
"logps/rejected": -226.1533966064453,
|
| 270 |
+
"loss": 0.6362,
|
| 271 |
+
"rewards/accuracies": 0.6640625,
|
| 272 |
+
"rewards/chosen": 0.014816783368587494,
|
| 273 |
+
"rewards/margins": 0.14678119122982025,
|
| 274 |
+
"rewards/rejected": -0.13196441531181335,
|
| 275 |
+
"step": 180
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.78,
|
| 279 |
+
"learning_rate": 4.1041347626339966e-07,
|
| 280 |
+
"logits/chosen": -2.447838306427002,
|
| 281 |
+
"logits/rejected": -2.4064364433288574,
|
| 282 |
+
"logps/chosen": -261.65130615234375,
|
| 283 |
+
"logps/rejected": -213.19497680664062,
|
| 284 |
+
"loss": 0.6274,
|
| 285 |
+
"rewards/accuracies": 0.653124988079071,
|
| 286 |
+
"rewards/chosen": 0.017185209318995476,
|
| 287 |
+
"rewards/margins": 0.16870170831680298,
|
| 288 |
+
"rewards/rejected": -0.15151652693748474,
|
| 289 |
+
"step": 190
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.83,
|
| 293 |
+
"learning_rate": 4.027565084226646e-07,
|
| 294 |
+
"logits/chosen": -2.471174716949463,
|
| 295 |
+
"logits/rejected": -2.4145748615264893,
|
| 296 |
+
"logps/chosen": -262.2930603027344,
|
| 297 |
+
"logps/rejected": -218.0220489501953,
|
| 298 |
+
"loss": 0.6186,
|
| 299 |
+
"rewards/accuracies": 0.6968749761581421,
|
| 300 |
+
"rewards/chosen": 0.027294564992189407,
|
| 301 |
+
"rewards/margins": 0.19551445543766022,
|
| 302 |
+
"rewards/rejected": -0.16821987926959991,
|
| 303 |
+
"step": 200
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.87,
|
| 307 |
+
"learning_rate": 3.9509954058192954e-07,
|
| 308 |
+
"logits/chosen": -2.4748778343200684,
|
| 309 |
+
"logits/rejected": -2.4354960918426514,
|
| 310 |
+
"logps/chosen": -283.8720703125,
|
| 311 |
+
"logps/rejected": -231.70785522460938,
|
| 312 |
+
"loss": 0.6158,
|
| 313 |
+
"rewards/accuracies": 0.6640625,
|
| 314 |
+
"rewards/chosen": 0.028042469173669815,
|
| 315 |
+
"rewards/margins": 0.21716149151325226,
|
| 316 |
+
"rewards/rejected": -0.18911901116371155,
|
| 317 |
+
"step": 210
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.91,
|
| 321 |
+
"learning_rate": 3.874425727411945e-07,
|
| 322 |
+
"logits/chosen": -2.4220242500305176,
|
| 323 |
+
"logits/rejected": -2.4089436531066895,
|
| 324 |
+
"logps/chosen": -276.76727294921875,
|
| 325 |
+
"logps/rejected": -229.3019561767578,
|
| 326 |
+
"loss": 0.6139,
|
| 327 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 328 |
+
"rewards/chosen": 0.01833787001669407,
|
| 329 |
+
"rewards/margins": 0.22341260313987732,
|
| 330 |
+
"rewards/rejected": -0.205074742436409,
|
| 331 |
+
"step": 220
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.95,
|
| 335 |
+
"learning_rate": 3.797856049004594e-07,
|
| 336 |
+
"logits/chosen": -2.4443860054016113,
|
| 337 |
+
"logits/rejected": -2.388517141342163,
|
| 338 |
+
"logps/chosen": -264.1051025390625,
|
| 339 |
+
"logps/rejected": -228.3242950439453,
|
| 340 |
+
"loss": 0.612,
|
| 341 |
+
"rewards/accuracies": 0.671875,
|
| 342 |
+
"rewards/chosen": 0.02690083347260952,
|
| 343 |
+
"rewards/margins": 0.2188110649585724,
|
| 344 |
+
"rewards/rejected": -0.19191020727157593,
|
| 345 |
+
"step": 230
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.99,
|
| 349 |
+
"learning_rate": 3.7212863705972436e-07,
|
| 350 |
+
"logits/chosen": -2.463099479675293,
|
| 351 |
+
"logits/rejected": -2.418534755706787,
|
| 352 |
+
"logps/chosen": -271.4806213378906,
|
| 353 |
+
"logps/rejected": -222.47207641601562,
|
| 354 |
+
"loss": 0.6123,
|
| 355 |
+
"rewards/accuracies": 0.6953125,
|
| 356 |
+
"rewards/chosen": 0.020688241347670555,
|
| 357 |
+
"rewards/margins": 0.21067467331886292,
|
| 358 |
+
"rewards/rejected": -0.1899864375591278,
|
| 359 |
+
"step": 240
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 1.0,
|
| 363 |
+
"eval_logits/chosen": -2.143613815307617,
|
| 364 |
+
"eval_logits/rejected": -2.023483991622925,
|
| 365 |
+
"eval_logps/chosen": -264.629150390625,
|
| 366 |
+
"eval_logps/rejected": -221.70089721679688,
|
| 367 |
+
"eval_loss": 0.6065443754196167,
|
| 368 |
+
"eval_rewards/accuracies": 0.6660000085830688,
|
| 369 |
+
"eval_rewards/chosen": 0.00029431533766910434,
|
| 370 |
+
"eval_rewards/margins": 0.2420923262834549,
|
| 371 |
+
"eval_rewards/rejected": -0.2417980283498764,
|
| 372 |
+
"eval_runtime": 601.9951,
|
| 373 |
+
"eval_samples_per_second": 3.322,
|
| 374 |
+
"eval_steps_per_second": 0.208,
|
| 375 |
+
"step": 242
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 1.03,
|
| 379 |
+
"learning_rate": 3.6447166921898925e-07,
|
| 380 |
+
"logits/chosen": -2.4006145000457764,
|
| 381 |
+
"logits/rejected": -2.347562789916992,
|
| 382 |
+
"logps/chosen": -257.4266052246094,
|
| 383 |
+
"logps/rejected": -210.4308319091797,
|
| 384 |
+
"loss": 0.6169,
|
| 385 |
+
"rewards/accuracies": 0.6703125238418579,
|
| 386 |
+
"rewards/chosen": 0.007514593191444874,
|
| 387 |
+
"rewards/margins": 0.2254376858472824,
|
| 388 |
+
"rewards/rejected": -0.21792307496070862,
|
| 389 |
+
"step": 250
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 1.07,
|
| 393 |
+
"learning_rate": 3.568147013782542e-07,
|
| 394 |
+
"logits/chosen": -2.4155120849609375,
|
| 395 |
+
"logits/rejected": -2.3756167888641357,
|
| 396 |
+
"logps/chosen": -261.9485778808594,
|
| 397 |
+
"logps/rejected": -226.1790313720703,
|
| 398 |
+
"loss": 0.5988,
|
| 399 |
+
"rewards/accuracies": 0.729687511920929,
|
| 400 |
+
"rewards/chosen": 0.02337762340903282,
|
| 401 |
+
"rewards/margins": 0.2698908746242523,
|
| 402 |
+
"rewards/rejected": -0.2465132772922516,
|
| 403 |
+
"step": 260
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 1.12,
|
| 407 |
+
"learning_rate": 3.4915773353751913e-07,
|
| 408 |
+
"logits/chosen": -2.429805278778076,
|
| 409 |
+
"logits/rejected": -2.366258144378662,
|
| 410 |
+
"logps/chosen": -278.4192810058594,
|
| 411 |
+
"logps/rejected": -236.13729858398438,
|
| 412 |
+
"loss": 0.5918,
|
| 413 |
+
"rewards/accuracies": 0.754687488079071,
|
| 414 |
+
"rewards/chosen": 0.02766202948987484,
|
| 415 |
+
"rewards/margins": 0.34231704473495483,
|
| 416 |
+
"rewards/rejected": -0.31465503573417664,
|
| 417 |
+
"step": 270
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 1.16,
|
| 421 |
+
"learning_rate": 3.41500765696784e-07,
|
| 422 |
+
"logits/chosen": -2.4361157417297363,
|
| 423 |
+
"logits/rejected": -2.3966078758239746,
|
| 424 |
+
"logps/chosen": -251.09671020507812,
|
| 425 |
+
"logps/rejected": -225.8013458251953,
|
| 426 |
+
"loss": 0.6033,
|
| 427 |
+
"rewards/accuracies": 0.7046874761581421,
|
| 428 |
+
"rewards/chosen": 0.009674707427620888,
|
| 429 |
+
"rewards/margins": 0.24585363268852234,
|
| 430 |
+
"rewards/rejected": -0.236178919672966,
|
| 431 |
+
"step": 280
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 1.2,
|
| 435 |
+
"learning_rate": 3.33843797856049e-07,
|
| 436 |
+
"logits/chosen": -2.408876895904541,
|
| 437 |
+
"logits/rejected": -2.394926071166992,
|
| 438 |
+
"logps/chosen": -283.1513671875,
|
| 439 |
+
"logps/rejected": -228.31497192382812,
|
| 440 |
+
"loss": 0.5912,
|
| 441 |
+
"rewards/accuracies": 0.7203124761581421,
|
| 442 |
+
"rewards/chosen": 0.0072790393605828285,
|
| 443 |
+
"rewards/margins": 0.3071500062942505,
|
| 444 |
+
"rewards/rejected": -0.29987096786499023,
|
| 445 |
+
"step": 290
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 1.24,
|
| 449 |
+
"learning_rate": 3.2618683001531396e-07,
|
| 450 |
+
"logits/chosen": -2.4088923931121826,
|
| 451 |
+
"logits/rejected": -2.3377671241760254,
|
| 452 |
+
"logps/chosen": -261.37591552734375,
|
| 453 |
+
"logps/rejected": -227.7656707763672,
|
| 454 |
+
"loss": 0.5887,
|
| 455 |
+
"rewards/accuracies": 0.7046874761581421,
|
| 456 |
+
"rewards/chosen": 0.010361125692725182,
|
| 457 |
+
"rewards/margins": 0.30056527256965637,
|
| 458 |
+
"rewards/rejected": -0.29020413756370544,
|
| 459 |
+
"step": 300
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.28,
|
| 463 |
+
"learning_rate": 3.1852986217457885e-07,
|
| 464 |
+
"logits/chosen": -2.457613468170166,
|
| 465 |
+
"logits/rejected": -2.3739213943481445,
|
| 466 |
+
"logps/chosen": -261.27642822265625,
|
| 467 |
+
"logps/rejected": -228.55892944335938,
|
| 468 |
+
"loss": 0.5963,
|
| 469 |
+
"rewards/accuracies": 0.7046874761581421,
|
| 470 |
+
"rewards/chosen": 0.00804700143635273,
|
| 471 |
+
"rewards/margins": 0.3003460764884949,
|
| 472 |
+
"rewards/rejected": -0.2922991216182709,
|
| 473 |
+
"step": 310
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.32,
|
| 477 |
+
"learning_rate": 3.108728943338438e-07,
|
| 478 |
+
"logits/chosen": -2.444277763366699,
|
| 479 |
+
"logits/rejected": -2.4352035522460938,
|
| 480 |
+
"logps/chosen": -252.82119750976562,
|
| 481 |
+
"logps/rejected": -237.87826538085938,
|
| 482 |
+
"loss": 0.5957,
|
| 483 |
+
"rewards/accuracies": 0.6796875,
|
| 484 |
+
"rewards/chosen": -0.017001762986183167,
|
| 485 |
+
"rewards/margins": 0.2523428499698639,
|
| 486 |
+
"rewards/rejected": -0.26934462785720825,
|
| 487 |
+
"step": 320
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 1.36,
|
| 491 |
+
"learning_rate": 3.0321592649310873e-07,
|
| 492 |
+
"logits/chosen": -2.424394130706787,
|
| 493 |
+
"logits/rejected": -2.357229232788086,
|
| 494 |
+
"logps/chosen": -253.7223358154297,
|
| 495 |
+
"logps/rejected": -224.33401489257812,
|
| 496 |
+
"loss": 0.5855,
|
| 497 |
+
"rewards/accuracies": 0.6953125,
|
| 498 |
+
"rewards/chosen": -0.0046251388266682625,
|
| 499 |
+
"rewards/margins": 0.3039107620716095,
|
| 500 |
+
"rewards/rejected": -0.3085358738899231,
|
| 501 |
+
"step": 330
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.4,
|
| 505 |
+
"learning_rate": 2.955589586523736e-07,
|
| 506 |
+
"logits/chosen": -2.4427707195281982,
|
| 507 |
+
"logits/rejected": -2.3827157020568848,
|
| 508 |
+
"logps/chosen": -265.708984375,
|
| 509 |
+
"logps/rejected": -226.42001342773438,
|
| 510 |
+
"loss": 0.5926,
|
| 511 |
+
"rewards/accuracies": 0.7046874761581421,
|
| 512 |
+
"rewards/chosen": -0.01725461333990097,
|
| 513 |
+
"rewards/margins": 0.3239946961402893,
|
| 514 |
+
"rewards/rejected": -0.3412492871284485,
|
| 515 |
+
"step": 340
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.45,
|
| 519 |
+
"learning_rate": 2.8790199081163856e-07,
|
| 520 |
+
"logits/chosen": -2.439563035964966,
|
| 521 |
+
"logits/rejected": -2.369589328765869,
|
| 522 |
+
"logps/chosen": -266.0650329589844,
|
| 523 |
+
"logps/rejected": -225.261474609375,
|
| 524 |
+
"loss": 0.5934,
|
| 525 |
+
"rewards/accuracies": 0.6640625,
|
| 526 |
+
"rewards/chosen": -0.026776671409606934,
|
| 527 |
+
"rewards/margins": 0.28162795305252075,
|
| 528 |
+
"rewards/rejected": -0.3084046244621277,
|
| 529 |
+
"step": 350
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.49,
|
| 533 |
+
"learning_rate": 2.802450229709035e-07,
|
| 534 |
+
"logits/chosen": -2.3998467922210693,
|
| 535 |
+
"logits/rejected": -2.349297523498535,
|
| 536 |
+
"logps/chosen": -254.8881378173828,
|
| 537 |
+
"logps/rejected": -213.3016815185547,
|
| 538 |
+
"loss": 0.5844,
|
| 539 |
+
"rewards/accuracies": 0.676562488079071,
|
| 540 |
+
"rewards/chosen": -0.03948161005973816,
|
| 541 |
+
"rewards/margins": 0.2969672679901123,
|
| 542 |
+
"rewards/rejected": -0.3364488482475281,
|
| 543 |
+
"step": 360
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.53,
|
| 547 |
+
"learning_rate": 2.725880551301684e-07,
|
| 548 |
+
"logits/chosen": -2.462137222290039,
|
| 549 |
+
"logits/rejected": -2.406816005706787,
|
| 550 |
+
"logps/chosen": -274.73101806640625,
|
| 551 |
+
"logps/rejected": -232.8585968017578,
|
| 552 |
+
"loss": 0.5927,
|
| 553 |
+
"rewards/accuracies": 0.692187488079071,
|
| 554 |
+
"rewards/chosen": -0.022728387266397476,
|
| 555 |
+
"rewards/margins": 0.31267625093460083,
|
| 556 |
+
"rewards/rejected": -0.3354046940803528,
|
| 557 |
+
"step": 370
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.57,
|
| 561 |
+
"learning_rate": 2.649310872894334e-07,
|
| 562 |
+
"logits/chosen": -2.448570728302002,
|
| 563 |
+
"logits/rejected": -2.4157679080963135,
|
| 564 |
+
"logps/chosen": -274.98968505859375,
|
| 565 |
+
"logps/rejected": -223.1295623779297,
|
| 566 |
+
"loss": 0.582,
|
| 567 |
+
"rewards/accuracies": 0.7281249761581421,
|
| 568 |
+
"rewards/chosen": -0.015962181612849236,
|
| 569 |
+
"rewards/margins": 0.37291616201400757,
|
| 570 |
+
"rewards/rejected": -0.38887840509414673,
|
| 571 |
+
"step": 380
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.61,
|
| 575 |
+
"learning_rate": 2.572741194486983e-07,
|
| 576 |
+
"logits/chosen": -2.4490561485290527,
|
| 577 |
+
"logits/rejected": -2.393211841583252,
|
| 578 |
+
"logps/chosen": -273.8176574707031,
|
| 579 |
+
"logps/rejected": -208.3898162841797,
|
| 580 |
+
"loss": 0.58,
|
| 581 |
+
"rewards/accuracies": 0.6968749761581421,
|
| 582 |
+
"rewards/chosen": -0.02870849333703518,
|
| 583 |
+
"rewards/margins": 0.3663511276245117,
|
| 584 |
+
"rewards/rejected": -0.39505964517593384,
|
| 585 |
+
"step": 390
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.65,
|
| 589 |
+
"learning_rate": 2.496171516079632e-07,
|
| 590 |
+
"logits/chosen": -2.465646266937256,
|
| 591 |
+
"logits/rejected": -2.399637222290039,
|
| 592 |
+
"logps/chosen": -293.25164794921875,
|
| 593 |
+
"logps/rejected": -239.49093627929688,
|
| 594 |
+
"loss": 0.5819,
|
| 595 |
+
"rewards/accuracies": 0.7328125238418579,
|
| 596 |
+
"rewards/chosen": 0.001335096312686801,
|
| 597 |
+
"rewards/margins": 0.3930011987686157,
|
| 598 |
+
"rewards/rejected": -0.39166611433029175,
|
| 599 |
+
"step": 400
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 1.69,
|
| 603 |
+
"learning_rate": 2.4196018376722816e-07,
|
| 604 |
+
"logits/chosen": -2.4297971725463867,
|
| 605 |
+
"logits/rejected": -2.400850296020508,
|
| 606 |
+
"logps/chosen": -278.59759521484375,
|
| 607 |
+
"logps/rejected": -228.43798828125,
|
| 608 |
+
"loss": 0.5868,
|
| 609 |
+
"rewards/accuracies": 0.6796875,
|
| 610 |
+
"rewards/chosen": -0.027199868112802505,
|
| 611 |
+
"rewards/margins": 0.3639487326145172,
|
| 612 |
+
"rewards/rejected": -0.3911486268043518,
|
| 613 |
+
"step": 410
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 1.74,
|
| 617 |
+
"learning_rate": 2.343032159264931e-07,
|
| 618 |
+
"logits/chosen": -2.4039382934570312,
|
| 619 |
+
"logits/rejected": -2.333247661590576,
|
| 620 |
+
"logps/chosen": -268.90826416015625,
|
| 621 |
+
"logps/rejected": -224.38961791992188,
|
| 622 |
+
"loss": 0.5785,
|
| 623 |
+
"rewards/accuracies": 0.729687511920929,
|
| 624 |
+
"rewards/chosen": -0.022465692833065987,
|
| 625 |
+
"rewards/margins": 0.4325350821018219,
|
| 626 |
+
"rewards/rejected": -0.45500072836875916,
|
| 627 |
+
"step": 420
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 1.78,
|
| 631 |
+
"learning_rate": 2.26646248085758e-07,
|
| 632 |
+
"logits/chosen": -2.383586883544922,
|
| 633 |
+
"logits/rejected": -2.335700273513794,
|
| 634 |
+
"logps/chosen": -259.71624755859375,
|
| 635 |
+
"logps/rejected": -217.78988647460938,
|
| 636 |
+
"loss": 0.5739,
|
| 637 |
+
"rewards/accuracies": 0.7171875238418579,
|
| 638 |
+
"rewards/chosen": -0.042887382209300995,
|
| 639 |
+
"rewards/margins": 0.38991934061050415,
|
| 640 |
+
"rewards/rejected": -0.43280667066574097,
|
| 641 |
+
"step": 430
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 1.82,
|
| 645 |
+
"learning_rate": 2.1898928024502298e-07,
|
| 646 |
+
"logits/chosen": -2.4449055194854736,
|
| 647 |
+
"logits/rejected": -2.387500047683716,
|
| 648 |
+
"logps/chosen": -263.4512939453125,
|
| 649 |
+
"logps/rejected": -221.47128295898438,
|
| 650 |
+
"loss": 0.5744,
|
| 651 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 652 |
+
"rewards/chosen": -0.019188478589057922,
|
| 653 |
+
"rewards/margins": 0.3951462507247925,
|
| 654 |
+
"rewards/rejected": -0.4143346846103668,
|
| 655 |
+
"step": 440
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.86,
|
| 659 |
+
"learning_rate": 2.113323124042879e-07,
|
| 660 |
+
"logits/chosen": -2.428199291229248,
|
| 661 |
+
"logits/rejected": -2.391083240509033,
|
| 662 |
+
"logps/chosen": -271.2774658203125,
|
| 663 |
+
"logps/rejected": -231.4476776123047,
|
| 664 |
+
"loss": 0.5749,
|
| 665 |
+
"rewards/accuracies": 0.721875011920929,
|
| 666 |
+
"rewards/chosen": -0.019557902589440346,
|
| 667 |
+
"rewards/margins": 0.4145377278327942,
|
| 668 |
+
"rewards/rejected": -0.4340956211090088,
|
| 669 |
+
"step": 450
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 1.9,
|
| 673 |
+
"learning_rate": 2.036753445635528e-07,
|
| 674 |
+
"logits/chosen": -2.4364569187164307,
|
| 675 |
+
"logits/rejected": -2.415283203125,
|
| 676 |
+
"logps/chosen": -284.0417785644531,
|
| 677 |
+
"logps/rejected": -238.28244018554688,
|
| 678 |
+
"loss": 0.5686,
|
| 679 |
+
"rewards/accuracies": 0.706250011920929,
|
| 680 |
+
"rewards/chosen": -0.022761384025216103,
|
| 681 |
+
"rewards/margins": 0.4516824781894684,
|
| 682 |
+
"rewards/rejected": -0.4744439125061035,
|
| 683 |
+
"step": 460
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 1.94,
|
| 687 |
+
"learning_rate": 1.9601837672281775e-07,
|
| 688 |
+
"logits/chosen": -2.4187042713165283,
|
| 689 |
+
"logits/rejected": -2.3860645294189453,
|
| 690 |
+
"logps/chosen": -270.18487548828125,
|
| 691 |
+
"logps/rejected": -236.39755249023438,
|
| 692 |
+
"loss": 0.5765,
|
| 693 |
+
"rewards/accuracies": 0.7171875238418579,
|
| 694 |
+
"rewards/chosen": -0.06018770858645439,
|
| 695 |
+
"rewards/margins": 0.3767903447151184,
|
| 696 |
+
"rewards/rejected": -0.4369780421257019,
|
| 697 |
+
"step": 470
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 1.98,
|
| 701 |
+
"learning_rate": 1.883614088820827e-07,
|
| 702 |
+
"logits/chosen": -2.438513994216919,
|
| 703 |
+
"logits/rejected": -2.3922152519226074,
|
| 704 |
+
"logps/chosen": -268.73028564453125,
|
| 705 |
+
"logps/rejected": -224.90036010742188,
|
| 706 |
+
"loss": 0.5841,
|
| 707 |
+
"rewards/accuracies": 0.71875,
|
| 708 |
+
"rewards/chosen": -0.046872757375240326,
|
| 709 |
+
"rewards/margins": 0.38516464829444885,
|
| 710 |
+
"rewards/rejected": -0.4320374131202698,
|
| 711 |
+
"step": 480
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 2.0,
|
| 715 |
+
"eval_logits/chosen": -2.1272027492523193,
|
| 716 |
+
"eval_logits/rejected": -2.0063846111297607,
|
| 717 |
+
"eval_logps/chosen": -265.2384338378906,
|
| 718 |
+
"eval_logps/rejected": -223.98130798339844,
|
| 719 |
+
"eval_loss": 0.5733410716056824,
|
| 720 |
+
"eval_rewards/accuracies": 0.7039999961853027,
|
| 721 |
+
"eval_rewards/chosen": -0.060633424669504166,
|
| 722 |
+
"eval_rewards/margins": 0.40920668840408325,
|
| 723 |
+
"eval_rewards/rejected": -0.46984007954597473,
|
| 724 |
+
"eval_runtime": 600.9449,
|
| 725 |
+
"eval_samples_per_second": 3.328,
|
| 726 |
+
"eval_steps_per_second": 0.208,
|
| 727 |
+
"step": 484
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 2.02,
|
| 731 |
+
"learning_rate": 1.807044410413476e-07,
|
| 732 |
+
"logits/chosen": -2.41025972366333,
|
| 733 |
+
"logits/rejected": -2.3768038749694824,
|
| 734 |
+
"logps/chosen": -257.9417419433594,
|
| 735 |
+
"logps/rejected": -236.35751342773438,
|
| 736 |
+
"loss": 0.5774,
|
| 737 |
+
"rewards/accuracies": 0.707812488079071,
|
| 738 |
+
"rewards/chosen": -0.0522170290350914,
|
| 739 |
+
"rewards/margins": 0.39746755361557007,
|
| 740 |
+
"rewards/rejected": -0.4496845602989197,
|
| 741 |
+
"step": 490
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 2.07,
|
| 745 |
+
"learning_rate": 1.7304747320061255e-07,
|
| 746 |
+
"logits/chosen": -2.4075393676757812,
|
| 747 |
+
"logits/rejected": -2.403886556625366,
|
| 748 |
+
"logps/chosen": -263.6119689941406,
|
| 749 |
+
"logps/rejected": -230.6759796142578,
|
| 750 |
+
"loss": 0.5768,
|
| 751 |
+
"rewards/accuracies": 0.7046874761581421,
|
| 752 |
+
"rewards/chosen": -0.023285437375307083,
|
| 753 |
+
"rewards/margins": 0.3978777825832367,
|
| 754 |
+
"rewards/rejected": -0.42116326093673706,
|
| 755 |
+
"step": 500
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 2.11,
|
| 759 |
+
"learning_rate": 1.6539050535987747e-07,
|
| 760 |
+
"logits/chosen": -2.479966402053833,
|
| 761 |
+
"logits/rejected": -2.3705756664276123,
|
| 762 |
+
"logps/chosen": -270.12005615234375,
|
| 763 |
+
"logps/rejected": -225.0802459716797,
|
| 764 |
+
"loss": 0.5698,
|
| 765 |
+
"rewards/accuracies": 0.706250011920929,
|
| 766 |
+
"rewards/chosen": -0.03792610391974449,
|
| 767 |
+
"rewards/margins": 0.43896961212158203,
|
| 768 |
+
"rewards/rejected": -0.4768957197666168,
|
| 769 |
+
"step": 510
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 2.15,
|
| 773 |
+
"learning_rate": 1.5773353751914243e-07,
|
| 774 |
+
"logits/chosen": -2.486560821533203,
|
| 775 |
+
"logits/rejected": -2.425957679748535,
|
| 776 |
+
"logps/chosen": -284.8712158203125,
|
| 777 |
+
"logps/rejected": -229.9952392578125,
|
| 778 |
+
"loss": 0.5714,
|
| 779 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 780 |
+
"rewards/chosen": -0.024752041324973106,
|
| 781 |
+
"rewards/margins": 0.4690770208835602,
|
| 782 |
+
"rewards/rejected": -0.49382907152175903,
|
| 783 |
+
"step": 520
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 2.19,
|
| 787 |
+
"learning_rate": 1.5007656967840735e-07,
|
| 788 |
+
"logits/chosen": -2.383312702178955,
|
| 789 |
+
"logits/rejected": -2.342586040496826,
|
| 790 |
+
"logps/chosen": -254.04122924804688,
|
| 791 |
+
"logps/rejected": -230.623046875,
|
| 792 |
+
"loss": 0.5678,
|
| 793 |
+
"rewards/accuracies": 0.7109375,
|
| 794 |
+
"rewards/chosen": -0.04037480801343918,
|
| 795 |
+
"rewards/margins": 0.4469054341316223,
|
| 796 |
+
"rewards/rejected": -0.4872801899909973,
|
| 797 |
+
"step": 530
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 2.23,
|
| 801 |
+
"learning_rate": 1.4241960183767226e-07,
|
| 802 |
+
"logits/chosen": -2.429459810256958,
|
| 803 |
+
"logits/rejected": -2.373309850692749,
|
| 804 |
+
"logps/chosen": -282.88043212890625,
|
| 805 |
+
"logps/rejected": -235.9270477294922,
|
| 806 |
+
"loss": 0.5726,
|
| 807 |
+
"rewards/accuracies": 0.703125,
|
| 808 |
+
"rewards/chosen": -0.0583997443318367,
|
| 809 |
+
"rewards/margins": 0.42233991622924805,
|
| 810 |
+
"rewards/rejected": -0.48073965311050415,
|
| 811 |
+
"step": 540
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 2.27,
|
| 815 |
+
"learning_rate": 1.347626339969372e-07,
|
| 816 |
+
"logits/chosen": -2.423274517059326,
|
| 817 |
+
"logits/rejected": -2.3877205848693848,
|
| 818 |
+
"logps/chosen": -270.8163757324219,
|
| 819 |
+
"logps/rejected": -242.1139678955078,
|
| 820 |
+
"loss": 0.5754,
|
| 821 |
+
"rewards/accuracies": 0.706250011920929,
|
| 822 |
+
"rewards/chosen": -0.015940938144922256,
|
| 823 |
+
"rewards/margins": 0.41127151250839233,
|
| 824 |
+
"rewards/rejected": -0.4272124767303467,
|
| 825 |
+
"step": 550
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 2.31,
|
| 829 |
+
"learning_rate": 1.2710566615620215e-07,
|
| 830 |
+
"logits/chosen": -2.3739233016967773,
|
| 831 |
+
"logits/rejected": -2.328768253326416,
|
| 832 |
+
"logps/chosen": -274.34735107421875,
|
| 833 |
+
"logps/rejected": -225.1607208251953,
|
| 834 |
+
"loss": 0.5587,
|
| 835 |
+
"rewards/accuracies": 0.7109375,
|
| 836 |
+
"rewards/chosen": -0.035870593041181564,
|
| 837 |
+
"rewards/margins": 0.445881187915802,
|
| 838 |
+
"rewards/rejected": -0.48175176978111267,
|
| 839 |
+
"step": 560
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 2.35,
|
| 843 |
+
"learning_rate": 1.1944869831546706e-07,
|
| 844 |
+
"logits/chosen": -2.399712085723877,
|
| 845 |
+
"logits/rejected": -2.379559278488159,
|
| 846 |
+
"logps/chosen": -266.9978332519531,
|
| 847 |
+
"logps/rejected": -238.74691772460938,
|
| 848 |
+
"loss": 0.5713,
|
| 849 |
+
"rewards/accuracies": 0.692187488079071,
|
| 850 |
+
"rewards/chosen": -0.07225313037633896,
|
| 851 |
+
"rewards/margins": 0.4113592207431793,
|
| 852 |
+
"rewards/rejected": -0.4836123585700989,
|
| 853 |
+
"step": 570
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 2.4,
|
| 857 |
+
"learning_rate": 1.11791730474732e-07,
|
| 858 |
+
"logits/chosen": -2.413283586502075,
|
| 859 |
+
"logits/rejected": -2.3746399879455566,
|
| 860 |
+
"logps/chosen": -262.73028564453125,
|
| 861 |
+
"logps/rejected": -226.4913787841797,
|
| 862 |
+
"loss": 0.5653,
|
| 863 |
+
"rewards/accuracies": 0.745312511920929,
|
| 864 |
+
"rewards/chosen": -0.06956593692302704,
|
| 865 |
+
"rewards/margins": 0.4847482740879059,
|
| 866 |
+
"rewards/rejected": -0.554314136505127,
|
| 867 |
+
"step": 580
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 2.44,
|
| 871 |
+
"learning_rate": 1.0413476263399694e-07,
|
| 872 |
+
"logits/chosen": -2.451280355453491,
|
| 873 |
+
"logits/rejected": -2.4071993827819824,
|
| 874 |
+
"logps/chosen": -269.5108947753906,
|
| 875 |
+
"logps/rejected": -219.07644653320312,
|
| 876 |
+
"loss": 0.5672,
|
| 877 |
+
"rewards/accuracies": 0.7281249761581421,
|
| 878 |
+
"rewards/chosen": -0.07316569238901138,
|
| 879 |
+
"rewards/margins": 0.4291161596775055,
|
| 880 |
+
"rewards/rejected": -0.5022818446159363,
|
| 881 |
+
"step": 590
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.48,
|
| 885 |
+
"learning_rate": 9.647779479326186e-08,
|
| 886 |
+
"logits/chosen": -2.393155097961426,
|
| 887 |
+
"logits/rejected": -2.3883702754974365,
|
| 888 |
+
"logps/chosen": -251.05142211914062,
|
| 889 |
+
"logps/rejected": -224.50564575195312,
|
| 890 |
+
"loss": 0.5763,
|
| 891 |
+
"rewards/accuracies": 0.6781250238418579,
|
| 892 |
+
"rewards/chosen": -0.06612516194581985,
|
| 893 |
+
"rewards/margins": 0.3570297658443451,
|
| 894 |
+
"rewards/rejected": -0.4231549799442291,
|
| 895 |
+
"step": 600
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 2.52,
|
| 899 |
+
"learning_rate": 8.88208269525268e-08,
|
| 900 |
+
"logits/chosen": -2.3813252449035645,
|
| 901 |
+
"logits/rejected": -2.39128041267395,
|
| 902 |
+
"logps/chosen": -260.7297058105469,
|
| 903 |
+
"logps/rejected": -223.4501495361328,
|
| 904 |
+
"loss": 0.5656,
|
| 905 |
+
"rewards/accuracies": 0.7109375,
|
| 906 |
+
"rewards/chosen": -0.07072791457176208,
|
| 907 |
+
"rewards/margins": 0.41946038603782654,
|
| 908 |
+
"rewards/rejected": -0.49018827080726624,
|
| 909 |
+
"step": 610
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.56,
|
| 913 |
+
"learning_rate": 8.116385911179173e-08,
|
| 914 |
+
"logits/chosen": -2.371321201324463,
|
| 915 |
+
"logits/rejected": -2.3626811504364014,
|
| 916 |
+
"logps/chosen": -279.528564453125,
|
| 917 |
+
"logps/rejected": -219.47598266601562,
|
| 918 |
+
"loss": 0.5688,
|
| 919 |
+
"rewards/accuracies": 0.723437488079071,
|
| 920 |
+
"rewards/chosen": -0.020319191738963127,
|
| 921 |
+
"rewards/margins": 0.4656984210014343,
|
| 922 |
+
"rewards/rejected": -0.4860176146030426,
|
| 923 |
+
"step": 620
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 2.6,
|
| 927 |
+
"learning_rate": 7.350689127105667e-08,
|
| 928 |
+
"logits/chosen": -2.4530327320098877,
|
| 929 |
+
"logits/rejected": -2.3786721229553223,
|
| 930 |
+
"logps/chosen": -272.66961669921875,
|
| 931 |
+
"logps/rejected": -232.73129272460938,
|
| 932 |
+
"loss": 0.5604,
|
| 933 |
+
"rewards/accuracies": 0.7015625238418579,
|
| 934 |
+
"rewards/chosen": -0.04867444932460785,
|
| 935 |
+
"rewards/margins": 0.4719608724117279,
|
| 936 |
+
"rewards/rejected": -0.5206353068351746,
|
| 937 |
+
"step": 630
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 2.64,
|
| 941 |
+
"learning_rate": 6.584992343032159e-08,
|
| 942 |
+
"logits/chosen": -2.3559648990631104,
|
| 943 |
+
"logits/rejected": -2.342261552810669,
|
| 944 |
+
"logps/chosen": -263.22344970703125,
|
| 945 |
+
"logps/rejected": -229.0948028564453,
|
| 946 |
+
"loss": 0.5758,
|
| 947 |
+
"rewards/accuracies": 0.6875,
|
| 948 |
+
"rewards/chosen": -0.0527966246008873,
|
| 949 |
+
"rewards/margins": 0.4219232201576233,
|
| 950 |
+
"rewards/rejected": -0.4747198522090912,
|
| 951 |
+
"step": 640
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 2.69,
|
| 955 |
+
"learning_rate": 5.819295558958652e-08,
|
| 956 |
+
"logits/chosen": -2.3997654914855957,
|
| 957 |
+
"logits/rejected": -2.3493194580078125,
|
| 958 |
+
"logps/chosen": -288.34991455078125,
|
| 959 |
+
"logps/rejected": -222.42855834960938,
|
| 960 |
+
"loss": 0.5643,
|
| 961 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 962 |
+
"rewards/chosen": -0.02724199928343296,
|
| 963 |
+
"rewards/margins": 0.5219191908836365,
|
| 964 |
+
"rewards/rejected": -0.5491611361503601,
|
| 965 |
+
"step": 650
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 2.73,
|
| 969 |
+
"learning_rate": 5.0535987748851455e-08,
|
| 970 |
+
"logits/chosen": -2.46504282951355,
|
| 971 |
+
"logits/rejected": -2.412127733230591,
|
| 972 |
+
"logps/chosen": -275.7881774902344,
|
| 973 |
+
"logps/rejected": -231.99331665039062,
|
| 974 |
+
"loss": 0.5616,
|
| 975 |
+
"rewards/accuracies": 0.7109375,
|
| 976 |
+
"rewards/chosen": -0.07443860173225403,
|
| 977 |
+
"rewards/margins": 0.41263580322265625,
|
| 978 |
+
"rewards/rejected": -0.48707443475723267,
|
| 979 |
+
"step": 660
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 2.77,
|
| 983 |
+
"learning_rate": 4.287901990811638e-08,
|
| 984 |
+
"logits/chosen": -2.4429898262023926,
|
| 985 |
+
"logits/rejected": -2.356980800628662,
|
| 986 |
+
"logps/chosen": -268.50396728515625,
|
| 987 |
+
"logps/rejected": -244.95095825195312,
|
| 988 |
+
"loss": 0.5638,
|
| 989 |
+
"rewards/accuracies": 0.7015625238418579,
|
| 990 |
+
"rewards/chosen": -0.04781431332230568,
|
| 991 |
+
"rewards/margins": 0.4686294496059418,
|
| 992 |
+
"rewards/rejected": -0.5164437294006348,
|
| 993 |
+
"step": 670
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 2.81,
|
| 997 |
+
"learning_rate": 3.522205206738132e-08,
|
| 998 |
+
"logits/chosen": -2.407198905944824,
|
| 999 |
+
"logits/rejected": -2.3732337951660156,
|
| 1000 |
+
"logps/chosen": -275.3802185058594,
|
| 1001 |
+
"logps/rejected": -230.15750122070312,
|
| 1002 |
+
"loss": 0.5689,
|
| 1003 |
+
"rewards/accuracies": 0.7109375,
|
| 1004 |
+
"rewards/chosen": -0.0815020427107811,
|
| 1005 |
+
"rewards/margins": 0.4394635558128357,
|
| 1006 |
+
"rewards/rejected": -0.5209656953811646,
|
| 1007 |
+
"step": 680
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 2.85,
|
| 1011 |
+
"learning_rate": 2.7565084226646246e-08,
|
| 1012 |
+
"logits/chosen": -2.3888936042785645,
|
| 1013 |
+
"logits/rejected": -2.376246929168701,
|
| 1014 |
+
"logps/chosen": -264.2990417480469,
|
| 1015 |
+
"logps/rejected": -236.7455291748047,
|
| 1016 |
+
"loss": 0.5745,
|
| 1017 |
+
"rewards/accuracies": 0.714062511920929,
|
| 1018 |
+
"rewards/chosen": -0.04266131669282913,
|
| 1019 |
+
"rewards/margins": 0.4692384600639343,
|
| 1020 |
+
"rewards/rejected": -0.5118998289108276,
|
| 1021 |
+
"step": 690
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 2.89,
|
| 1025 |
+
"learning_rate": 1.9908116385911178e-08,
|
| 1026 |
+
"logits/chosen": -2.4064769744873047,
|
| 1027 |
+
"logits/rejected": -2.388720989227295,
|
| 1028 |
+
"logps/chosen": -260.6070556640625,
|
| 1029 |
+
"logps/rejected": -227.61105346679688,
|
| 1030 |
+
"loss": 0.5712,
|
| 1031 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1032 |
+
"rewards/chosen": -0.09105347096920013,
|
| 1033 |
+
"rewards/margins": 0.38188761472702026,
|
| 1034 |
+
"rewards/rejected": -0.4729411005973816,
|
| 1035 |
+
"step": 700
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.93,
|
| 1039 |
+
"learning_rate": 1.225114854517611e-08,
|
| 1040 |
+
"logits/chosen": -2.43709659576416,
|
| 1041 |
+
"logits/rejected": -2.371903657913208,
|
| 1042 |
+
"logps/chosen": -280.87945556640625,
|
| 1043 |
+
"logps/rejected": -221.88931274414062,
|
| 1044 |
+
"loss": 0.5684,
|
| 1045 |
+
"rewards/accuracies": 0.7171875238418579,
|
| 1046 |
+
"rewards/chosen": -0.06613589823246002,
|
| 1047 |
+
"rewards/margins": 0.4434446394443512,
|
| 1048 |
+
"rewards/rejected": -0.5095804929733276,
|
| 1049 |
+
"step": 710
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 2.97,
|
| 1053 |
+
"learning_rate": 4.594180704441042e-09,
|
| 1054 |
+
"logits/chosen": -2.400611639022827,
|
| 1055 |
+
"logits/rejected": -2.3782918453216553,
|
| 1056 |
+
"logps/chosen": -257.98419189453125,
|
| 1057 |
+
"logps/rejected": -225.0382843017578,
|
| 1058 |
+
"loss": 0.567,
|
| 1059 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1060 |
+
"rewards/chosen": -0.05685793235898018,
|
| 1061 |
+
"rewards/margins": 0.41515034437179565,
|
| 1062 |
+
"rewards/rejected": -0.47200828790664673,
|
| 1063 |
+
"step": 720
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 3.0,
|
| 1067 |
+
"eval_logits/chosen": -2.1211588382720947,
|
| 1068 |
+
"eval_logits/rejected": -2.000145435333252,
|
| 1069 |
+
"eval_logps/chosen": -265.46588134765625,
|
| 1070 |
+
"eval_logps/rejected": -224.6123809814453,
|
| 1071 |
+
"eval_loss": 0.5657259225845337,
|
| 1072 |
+
"eval_rewards/accuracies": 0.7020000219345093,
|
| 1073 |
+
"eval_rewards/chosen": -0.08337792754173279,
|
| 1074 |
+
"eval_rewards/margins": 0.44957080483436584,
|
| 1075 |
+
"eval_rewards/rejected": -0.5329487323760986,
|
| 1076 |
+
"eval_runtime": 600.5346,
|
| 1077 |
+
"eval_samples_per_second": 3.33,
|
| 1078 |
+
"eval_steps_per_second": 0.208,
|
| 1079 |
+
"step": 726
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 3.0,
|
| 1083 |
+
"step": 726,
|
| 1084 |
+
"total_flos": 0.0,
|
| 1085 |
+
"train_loss": 0.6034470564241908,
|
| 1086 |
+
"train_runtime": 84521.5132,
|
| 1087 |
+
"train_samples_per_second": 2.199,
|
| 1088 |
+
"train_steps_per_second": 0.009
|
| 1089 |
+
}
|
| 1090 |
+
],
|
| 1091 |
+
"logging_steps": 10,
|
| 1092 |
+
"max_steps": 726,
|
| 1093 |
+
"num_train_epochs": 3,
|
| 1094 |
+
"save_steps": 500,
|
| 1095 |
+
"total_flos": 0.0,
|
| 1096 |
+
"trial_name": null,
|
| 1097 |
+
"trial_params": null
|
| 1098 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2cc2d01d777cefbd752ab7c173c33376515874ab7b2ff1d04da2abc8678e0b2d
|
| 3 |
+
size 4283
|