chest2vec_0.6b_chest / chest2vec_config.json
lukeingawesome
Simplify model to global embeddings only (no section pooling), update adapter weights from resume3 checkpoint-11270
999f0c9
raw
history blame contribute delete
208 Bytes
{
"name": "chest2vec_0.6b_chest",
"base_model": "Qwen/Qwen3-Embedding-0.6B",
"adapter_subdir": "contrastive",
"require_flash_attention_2": true,
"default_max_len": 512,
"pooling": "last_token"
}