初始化项目,由ModelHub XC社区提供模型
Model: Omarrran/koshur-kouter-ks-en_v1 Source: Original Platform
This commit is contained in:
39
artifacts/config.json
Normal file
39
artifacts/config.json
Normal file
@@ -0,0 +1,39 @@
|
||||
{
|
||||
"model_id": "sarvamai/sarvam-translate",
|
||||
"max_seq_length": 512,
|
||||
"batch_size": 32,
|
||||
"grad_accum": 1,
|
||||
"use_bf16": true,
|
||||
"s1_lora_r": 32,
|
||||
"s1_lora_alpha": 64,
|
||||
"s1_lora_dropout": 0.05,
|
||||
"s1_lr": 0.0002,
|
||||
"s1_epochs": 2,
|
||||
"s1_warmup_ratio": 0.03,
|
||||
"s2_lora_r": 16,
|
||||
"s2_lora_alpha": 32,
|
||||
"s2_lora_dropout": 0.05,
|
||||
"s2_lr": 5e-05,
|
||||
"s2_epochs": 3,
|
||||
"s2_warmup_ratio": 0.05,
|
||||
"save_steps": 250,
|
||||
"eval_steps": 250,
|
||||
"logging_steps": 10,
|
||||
"lr_scheduler": "cosine",
|
||||
"optimizer": "paged_adamw_8bit",
|
||||
"weight_decay": 0.01,
|
||||
"max_grad_norm": 1.0,
|
||||
"lora_targets": [
|
||||
"q_proj",
|
||||
"k_proj",
|
||||
"v_proj",
|
||||
"o_proj",
|
||||
"gate_proj",
|
||||
"up_proj",
|
||||
"down_proj"
|
||||
],
|
||||
"modules_to_save": [
|
||||
"lm_head",
|
||||
"embed_tokens"
|
||||
]
|
||||
}
|
||||
Reference in New Issue
Block a user