| { | |
| "_name_or_path": "/project2/lichanghao_w/lch/retrain-heyna/checkpoint-15000", | |
| "activation_freq": 10, | |
| "architectures": [ | |
| "PlantGLMForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_plantglm.PlantGLMConfig", | |
| "AutoModel": "modeling_plantglm.PlantGLMModel", | |
| "AutoModelForCausalLM": "modeling_plantglm.PlantGLMForCausalLM", | |
| "AutoModelForSequenceClassification": "modeling_plantglm.PlantGLMForSequenceClassification" | |
| }, | |
| "d_inner": 3072, | |
| "d_model": 1024, | |
| "emb_dim": 9, | |
| "embed_dropout": 0.1, | |
| "filter_order": 256, | |
| "hyena_dropout": 0.0, | |
| "hyena_filter_dropout": 0.0, | |
| "hyena_order": 2, | |
| "initializer_range": 0.02, | |
| "max_seq_len": 65538, | |
| "model_type": "plantglm", | |
| "n_layer": 16, | |
| "num_inner_mlps": 2, | |
| "pad_token_id": 0, | |
| "pad_vocab_size_multiple": 8, | |
| "rms_norm_epsilon": 1e-06, | |
| "short_filter_order": 3, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "float32", | |
| "train_freq": true, | |
| "transformers_version": "4.44.2", | |
| "use_bias": true, | |
| "vocab_size": 24 | |
| } | |