| { | |
| "architectures": [ | |
| "TorchMultiOmicsModel" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "chatNT.ChatNTConfig", | |
| "AutoModel": "chatNT.TorchMultiOmicsModel" | |
| }, | |
| "bio_pad_token_id": 1, | |
| "english_pad_token_id": 2, | |
| "esm_config": { | |
| "add_bias_ffn": false, | |
| "add_bias_kv": false, | |
| "alphabet_size": 4107, | |
| "attention_heads": 16, | |
| "attention_maps_to_save": [], | |
| "bias_word_embedding": false, | |
| "emb_layer_norm_before": false, | |
| "embed_dim": 1024, | |
| "embed_scale": 1.0, | |
| "embeddings_layers_to_save": [ | |
| 21 | |
| ], | |
| "ffn_activation_name": "swish", | |
| "ffn_embed_dim": 4096, | |
| "key_size": 64, | |
| "layer_norm_eps": 1e-05, | |
| "lm_head": "roberta", | |
| "mask_before_attention": false, | |
| "mask_token_id": 2, | |
| "masking_prob": 0.0, | |
| "masking_ratio": 0.0, | |
| "max_positions": 2048, | |
| "num_layers": 29, | |
| "pad_token_id": 1, | |
| "positional_embedding": null, | |
| "pre_layer_norm": true, | |
| "rescaling_factor": null, | |
| "token_dropout": false, | |
| "use_glu_in_ffn": true, | |
| "use_gradient_checkpointing": false, | |
| "use_rotary_embedding": true | |
| }, | |
| "gpt_config": { | |
| "add_bias_attn": false, | |
| "add_bias_ffn": false, | |
| "add_bias_lm_head": false, | |
| "embed_dim": 4096, | |
| "eos_token_id": 2, | |
| "ffn_activation_name": "silu", | |
| "ffn_embed_dim": 11008, | |
| "norm_type": "RMS_norm", | |
| "num_heads": 32, | |
| "num_kv_heads": 32, | |
| "num_layers": 32, | |
| "parallel_attention_ff": false, | |
| "rms_norm_eps": 1e-06, | |
| "rope_config": { | |
| "dim": 128, | |
| "max_seq_len": 2048, | |
| "theta": 10000.0 | |
| }, | |
| "use_glu_in_ffn": true, | |
| "use_gradient_checkpointing": false, | |
| "vocab_size": 32000 | |
| }, | |
| "model_type": "ChatNT", | |
| "perceiver_resampler_config": { | |
| "add_bias_ffn": true, | |
| "add_bias_kv": false, | |
| "attention_heads": 32, | |
| "emb_layer_norm_before": false, | |
| "embed_dim": 4096, | |
| "ffn_activation_name": "gelu-no-approx", | |
| "ffn_embed_dim": 11008, | |
| "key_size": 128, | |
| "num_layers": 3, | |
| "resampled_length": 64, | |
| "use_glu_in_ffn": false, | |
| "use_gradient_checkpointing": false | |
| }, | |
| "seq_token_id": 32000, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.41.1" | |
| } | |