| { |
| "architectures": [ |
| "BltModel" |
| ], |
| "bos_token_id": null, |
| "cross_attn_k": 2, |
| "decoder_config": { |
| "bos_token_id": null, |
| "cross_attn_all_layers": true, |
| "cross_attn_k": 2, |
| "dropout": 0.0, |
| "eos_token_id": null, |
| "head_dim": 8, |
| "hidden_act": "silu", |
| "hidden_size": 16, |
| "hidden_size_global": 32, |
| "initializer_range": 0.02, |
| "intermediate_size": 32, |
| "max_position_embeddings": 32, |
| "model_type": "blt_local_decoder", |
| "num_attention_heads": 2, |
| "num_hidden_layers": 1, |
| "num_key_value_heads": 2, |
| "pad_token_id": null, |
| "rms_norm_eps": 1e-05, |
| "rope_parameters": { |
| "rope_theta": 500000.0, |
| "rope_type": "default" |
| }, |
| "tie_word_embeddings": false, |
| "vocab_size": 261 |
| }, |
| "dtype": "float32", |
| "encoder_config": { |
| "cross_attn_all_layers": false, |
| "cross_attn_k": 2, |
| "dropout": 0.0, |
| "hidden_act": "silu", |
| "hidden_size": 16, |
| "hidden_size_global": 2048, |
| "initializer_range": 0.02, |
| "intermediate_size": 32, |
| "max_position_embeddings": 32, |
| "model_type": "blt_local_encoder", |
| "num_attention_heads": 2, |
| "num_hidden_layers": 1, |
| "num_key_value_heads": 2, |
| "rms_norm_eps": 1e-05, |
| "rope_parameters": { |
| "rope_theta": 500000.0, |
| "rope_type": "default" |
| }, |
| "tie_word_embeddings": false, |
| "vocab_size": 261 |
| }, |
| "encoder_hash_byte_group_nb_functions": 1, |
| "encoder_hash_byte_group_size": [ |
| 2, |
| 3 |
| ], |
| "encoder_hash_byte_group_vocab": 64, |
| "eos_token_id": null, |
| "global_config": { |
| "dropout": 0.0, |
| "encoder_cross_output_size": null, |
| "head_dim": 16, |
| "hidden_act": "silu", |
| "hidden_size": 32, |
| "initializer_range": 0.02, |
| "intermediate_size": 32, |
| "max_position_embeddings": 32, |
| "model_type": "blt_global_transformer", |
| "num_attention_heads": 2, |
| "num_hidden_layers": 1, |
| "num_key_value_heads": 2, |
| "rms_norm_eps": 1e-05, |
| "rope_parameters": { |
| "rope_theta": 500000.0, |
| "rope_type": "default" |
| }, |
| "tie_word_embeddings": false |
| }, |
| "hidden_size": 16, |
| "initializer_range": 0.02, |
| "max_patch_length": null, |
| "max_position_embeddings": 32, |
| "model_type": "blt", |
| "num_attention_heads": 2, |
| "num_hidden_layers": 1, |
| "pad_token_id": null, |
| "patch_in_forward": false, |
| "patch_size": 4, |
| "patcher_config": { |
| "dropout": 0.0, |
| "head_dim": 8, |
| "hidden_act": "silu", |
| "hidden_size": 16, |
| "initializer_range": 0.02, |
| "intermediate_size": 32, |
| "max_position_embeddings": 32, |
| "model_type": "blt_patcher", |
| "num_attention_heads": 2, |
| "num_hidden_layers": 1, |
| "num_key_value_heads": 2, |
| "rms_norm_eps": 1e-05, |
| "rope_parameters": { |
| "rope_theta": 500000.0, |
| "rope_type": "default" |
| }, |
| "tie_word_embeddings": false, |
| "vocab_size": 261 |
| }, |
| "patching_batch_size": 1, |
| "patching_mode": "entropy", |
| "patching_threshold": 1.335442066192627, |
| "rope_parameters": { |
| "rope_theta": 500000.0, |
| "rope_type": "default" |
| }, |
| "tie_word_embeddings": false, |
| "transformers_version": "5.15.0.dev0", |
| "use_cache": true, |
| "vocab_size": 261 |
| } |
|
|