| { | |
| "architectures": [ | |
| "MUDDPythia" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_muddpythia.MUDDPythiaConfig", | |
| "AutoModelForCausalLM": "modeling_muddpythia.MUDDPythia" | |
| }, | |
| "block_size": 2048, | |
| "bos_token_id": 0, | |
| "dense": true, | |
| "dense_type": "qkvr", | |
| "dim": 2560, | |
| "dynamic_dense": true, | |
| "eos_token_id": 0, | |
| "expand_last": true, | |
| "head_dim": 80, | |
| "intermediate_size": 10240, | |
| "is_training": false, | |
| "model_type": "muddpythia", | |
| "n_head": 32, | |
| "n_layer": 32, | |
| "n_local_heads": 32, | |
| "norm_eps": 1e-05, | |
| "rope_base": 10000, | |
| "rotary_pct": 0.25, | |
| "round64": true, | |
| "sepln": true, | |
| "stack_hidden": false, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.35.0", | |
| "use_gradient_checkpointing": false, | |
| "use_layer_cache": true, | |
| "use_linear_bias": true, | |
| "use_parallel_residual": true, | |
| "use_qk_norm": true, | |
| "vocab_size": 50432 | |
| } | |