| { |
| "_name_or_path": "/run/media/knut/HD/huggingface models/language models/LLMs/pretrained/3b/btlm-3b-8k-base/", |
| "activation_function": "swiglu", |
| "architectures": [ |
| "BTLMLMHeadModel" |
| ], |
| "attn_pdrop": 0.0, |
| "auto_map": { |
| "AutoConfig": "configuration_btlm.BTLMConfig", |
| "AutoModel": "modeling_btlm.BTLMModel", |
| "AutoModelForCausalLM": "modeling_btlm.BTLMLMHeadModel", |
| "AutoModelForQuestionAnswering": "modeling_btlm.BTLMForQuestionAnswering", |
| "AutoModelForSequenceClassification": "modeling_btlm.BTLMForSequenceClassification", |
| "AutoModelForTokenClassification": "modeling_btlm.BTLMForTokenClassification" |
| }, |
| "bos_token_id": 50256, |
| "embd_pdrop": 0.0, |
| "eos_token_id": 50256, |
| "initializer_range": 0.073, |
| "layer_norm_epsilon": 1e-05, |
| "model_type": "btlm", |
| "mup_embeddings_scale": 14.6, |
| "mup_output_alpha": 2.22, |
| "mup_scale_qk_dot_by_d": true, |
| "mup_width_scale": 0.1, |
| "n_embd": 2560, |
| "n_head": 32, |
| "n_inner": 6826, |
| "n_layer": 32, |
| "n_positions": 8192, |
| "position_embedding_type": "alibi", |
| "reorder_and_upcast_attn": false, |
| "resid_pdrop": 0.0, |
| "scale_attn_by_inverse_layer_idx": false, |
| "scale_attn_weights": true, |
| "torch_dtype": "float16", |
| "transformers_version": "4.35.0", |
| "use_cache": true, |
| "vocab_size": 50257 |
| } |
|
|