| { | |
| "_name_or_path": "japanese-stablelm-kirisame-v3", | |
| "architectures": [ | |
| "JapaneseStableLMAlphaForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_japanese_stablelm_alpha.JapaneseStableLMAlphaConfig", | |
| "AutoModelForCausalLM": "modeling_japanese_stablelm_alpha.JapaneseStableLMAlphaForCausalLM" | |
| }, | |
| "bos_token_id": 3, | |
| "classifier_dropout": 0.1, | |
| "eos_token_id": 3, | |
| "hidden_act": "silu", | |
| "hidden_size": 4096, | |
| "initializer_range": 0.02, | |
| "layer_norm_eps": 1e-05, | |
| "max_position_embeddings": 2048, | |
| "num_attention_heads": 32, | |
| "num_hidden_layers": 32, | |
| "rotary_emb_base": 10000, | |
| "rotary_pct": 0.25, | |
| "rotary_scale_base": 512, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "float16", | |
| "transformers_version": "4.32.0.dev0", | |
| "use_bias_in_mlp": false, | |
| "use_cache": true, | |
| "use_parallel_residual": true, | |
| "vocab_size": 65536 | |
| } | |