初始化项目,由ModelHub XC社区提供模型
Model: Pomilon/LEMA-llama-2-7b Source: Original Platform
This commit is contained in:
3
final_checkpoint/adapter_model.bin
Normal file
3
final_checkpoint/adapter_model.bin
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:fd3a4aac8000ca9619a3b1e90588f3f8b13676b3d854a7332bba2179b83a8b3c
|
||||
size 160047395
|
||||
30
final_checkpoint/lema_config.json
Normal file
30
final_checkpoint/lema_config.json
Normal file
@@ -0,0 +1,30 @@
|
||||
{
|
||||
"model_name_or_path": "NousResearch/Llama-2-7b-hf",
|
||||
"model_type": null,
|
||||
"gbi_path": "llama2_7b.safetensors",
|
||||
"device": "cuda",
|
||||
"strategy": "streaming",
|
||||
"ram_buffer_size": 2,
|
||||
"vram_buffer_size": 1,
|
||||
"use_lora": true,
|
||||
"lora_rank": 16,
|
||||
"lora_alpha": 32,
|
||||
"lora_target_modules": [
|
||||
"q_proj",
|
||||
"k_proj",
|
||||
"v_proj",
|
||||
"o_proj",
|
||||
"gate_proj",
|
||||
"up_proj",
|
||||
"down_proj"
|
||||
],
|
||||
"learning_rate": 0.0001,
|
||||
"batch_size": 8,
|
||||
"gradient_accumulation_steps": 1,
|
||||
"max_seq_length": 512,
|
||||
"gradient_checkpointing": true,
|
||||
"save_steps": 500,
|
||||
"output_dir": "checkpoints",
|
||||
"dtype": "float16",
|
||||
"attn_implementation": "eager"
|
||||
}
|
||||
3
final_checkpoint/optimizer.bin
Normal file
3
final_checkpoint/optimizer.bin
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:ab96dae4a5527346060ce770b39e12b54c79dd9e80d37460f5c1f32b639fac78
|
||||
size 320194275
|
||||
Reference in New Issue
Block a user