| { | |
| "architectures": [ | |
| "TinyMixtralForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "modeling_tiny_mixtral.TinyMixtralConfig", | |
| "AutoModelForCausalLM": "modeling_tiny_mixtral.TinyMixtralForCausalLM" | |
| }, | |
| "attn_dropout": 0.0, | |
| "attn_eps": 1e-06, | |
| "d_head": 64, | |
| "d_model": 768, | |
| "device": "cpu", | |
| "dropout": 0.0, | |
| "ffn_eps": 1e-06, | |
| "max_seq_len": 1024, | |
| "model_type": "tiny_mixtral_5l_total", | |
| "n_experts": 8, | |
| "n_heads": 12, | |
| "n_layers": 5, | |
| "norm_eps": 1e-06, | |
| "top_k": null, | |
| "top_k_experts": 2, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.53.2", | |
| "vocab_size": 50257 | |
| } | |