| { | |
| "architectures": [ | |
| "DbrxForCausalLM" | |
| ], | |
| "attn_config": { | |
| "model_type": "" | |
| }, | |
| "d_model": 8, | |
| "emb_pdrop": 0.0, | |
| "ffn_config": { | |
| "model_type": "" | |
| }, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 32, | |
| "max_seq_len": 2048, | |
| "model_type": "dbrx", | |
| "n_heads": 4, | |
| "n_layers": 2, | |
| "num_key_value_heads": 2, | |
| "output_router_logits": false, | |
| "resid_pdrop": 0.0, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.47.0.dev0", | |
| "use_cache": true, | |
| "vocab_size": 100302 | |
| } | |