初始化项目,由ModelHub XC社区提供模型
Model: mims-harvard/bio-posttrain-qwen3-1.7b-dna-sft Source: Original Platform
This commit is contained in:
20
dna_model_config.json
Normal file
20
dna_model_config.json
Normal file
@@ -0,0 +1,20 @@
|
||||
{
|
||||
"model_type": "dna_llm",
|
||||
"dna_model_name": "evo2_1b_base",
|
||||
"dna_is_evo2": true,
|
||||
"dna_embedding_layer": "blocks.20.mlp.l3",
|
||||
"dna_hidden_size": 1920,
|
||||
"text_hidden_size": 2048,
|
||||
"max_length_dna": 2048,
|
||||
"max_length_text": 2048,
|
||||
"special_tokens": [
|
||||
"<|dna_start|>",
|
||||
"<|dna_pad|>",
|
||||
"<|dna_end|>"
|
||||
],
|
||||
"notes": "DNA encoder weights are not bundled. Load Evo2 evo2_1b_base separately and apply dna_projection.pt to map encoder embeddings into the text model.",
|
||||
"stage": "sft",
|
||||
"lora_rank": 64,
|
||||
"lora_alpha": 128,
|
||||
"val_loss": 0.4687
|
||||
}
|
||||
Reference in New Issue
Block a user