| { | |
| "_name_or_path": "/data/workspace/bat-gpt/checkpoints/batgpt-15b-sirius", | |
| "alibi": false, | |
| "architectures": [ | |
| "BatGPTForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_batgpt.BatGPTConfig", | |
| "AutoModel": "modeling_batgpt.BatGPTForCausalLM", | |
| "AutoModelForCausalLM": "modeling_batgpt.BatGPTForCausalLM", | |
| "AutoModelForSeq2SeqLM": "modeling_batgpt.BatGPTForCausalLM" | |
| }, | |
| "emb_dim": 5632, | |
| "empty_init": false, | |
| "eos_token_id": 2, | |
| "ffn_hidden_size": 13696, | |
| "hidden_dropout": 0.0, | |
| "hidden_size": 5632, | |
| "layer_norm_epsilon": 1e-05, | |
| "max_seq_len": 32768, | |
| "mlp_activation": "swiglu", | |
| "model_type": "batgpt", | |
| "n_head": 44, | |
| "n_layer": 48, | |
| "num_heads_per_kv": 2, | |
| "pad_token_id": 0, | |
| "pos_emb_impl": "rope", | |
| "prefix_proj": false, | |
| "prefix_size": null, | |
| "qkv_bias": true, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "float16", | |
| "transformers_version": "4.30.2", | |
| "use_cache": true, | |
| "use_emb_factorization": false, | |
| "use_multi_query_attn": true, | |
| "use_native_attn_impl": true, | |
| "vocab_size": 65536 | |
| } | |