Update config.json
Browse files- config.json +57 -35
config.json
CHANGED
|
@@ -1,37 +1,59 @@
|
|
| 1 |
{
|
| 2 |
-
"
|
| 3 |
-
"
|
| 4 |
-
|
| 5 |
-
"bias": "none",
|
| 6 |
-
"eva_config": null,
|
| 7 |
-
"exclude_modules": null,
|
| 8 |
-
"fan_in_fan_out": false,
|
| 9 |
-
"inference_mode": true,
|
| 10 |
-
"init_lora_weights": true,
|
| 11 |
-
"layer_replication": null,
|
| 12 |
-
"layers_pattern": null,
|
| 13 |
-
"layers_to_transform": null,
|
| 14 |
-
"loftq_config": {},
|
| 15 |
-
"lora_alpha": 16,
|
| 16 |
-
"lora_bias": false,
|
| 17 |
-
"lora_dropout": 0,
|
| 18 |
-
"megatron_config": null,
|
| 19 |
-
"megatron_core": "megatron.core",
|
| 20 |
-
"modules_to_save": null,
|
| 21 |
-
"peft_type": "LORA",
|
| 22 |
-
"r": 16,
|
| 23 |
-
"rank_pattern": {},
|
| 24 |
-
"revision": null,
|
| 25 |
-
"target_modules": [
|
| 26 |
-
"gate_proj",
|
| 27 |
-
"down_proj",
|
| 28 |
-
"o_proj",
|
| 29 |
-
"q_proj",
|
| 30 |
-
"up_proj",
|
| 31 |
-
"v_proj",
|
| 32 |
-
"k_proj"
|
| 33 |
],
|
| 34 |
-
"
|
| 35 |
-
"
|
| 36 |
-
"
|
| 37 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "meta-llama/Llama-3.2-3B-Instruct",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LlamaForCausalLM"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
],
|
| 6 |
+
"attention_bias": false,
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"bos_token_id": 128000,
|
| 9 |
+
"eos_token_id": 128009,
|
| 10 |
+
"head_dim": 128,
|
| 11 |
+
"hidden_act": "silu",
|
| 12 |
+
"hidden_size": 3072,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 8192,
|
| 15 |
+
"max_position_embeddings": 131072,
|
| 16 |
+
"mlp_bias": false,
|
| 17 |
+
"model_type": "llama",
|
| 18 |
+
"num_attention_heads": 24,
|
| 19 |
+
"num_hidden_layers": 28,
|
| 20 |
+
"num_key_value_heads": 8,
|
| 21 |
+
"pad_token_id": 128004,
|
| 22 |
+
"pretraining_tp": 1,
|
| 23 |
+
"quantization_config": {
|
| 24 |
+
"_load_in_4bit": true,
|
| 25 |
+
"_load_in_8bit": false,
|
| 26 |
+
"bnb_4bit_compute_dtype": "bfloat16",
|
| 27 |
+
"bnb_4bit_quant_storage": "uint8",
|
| 28 |
+
"bnb_4bit_quant_type": "nf4",
|
| 29 |
+
"bnb_4bit_use_double_quant": true,
|
| 30 |
+
"llm_int8_enable_fp32_cpu_offload": false,
|
| 31 |
+
"llm_int8_has_fp16_weight": false,
|
| 32 |
+
"llm_int8_skip_modules": [
|
| 33 |
+
"lm_head",
|
| 34 |
+
"multi_modal_projector",
|
| 35 |
+
"merger",
|
| 36 |
+
"modality_projection",
|
| 37 |
+
"model.layers.1.mlp"
|
| 38 |
+
],
|
| 39 |
+
"llm_int8_threshold": 6.0,
|
| 40 |
+
"load_in_4bit": true,
|
| 41 |
+
"load_in_8bit": false,
|
| 42 |
+
"quant_method": "bitsandbytes"
|
| 43 |
+
},
|
| 44 |
+
"rms_norm_eps": 1e-05,
|
| 45 |
+
"rope_scaling": {
|
| 46 |
+
"factor": 32.0,
|
| 47 |
+
"high_freq_factor": 4.0,
|
| 48 |
+
"low_freq_factor": 1.0,
|
| 49 |
+
"original_max_position_embeddings": 8192,
|
| 50 |
+
"rope_type": "llama3"
|
| 51 |
+
},
|
| 52 |
+
"rope_theta": 500000.0,
|
| 53 |
+
"tie_word_embeddings": true,
|
| 54 |
+
"torch_dtype": "bfloat16",
|
| 55 |
+
"transformers_version": "4.48.1",
|
| 56 |
+
"unsloth_fixed": true,
|
| 57 |
+
"use_cache": true,
|
| 58 |
+
"vocab_size": 128256
|
| 59 |
+
}
|