{ "_name_or_path": "checkpoints/Llama-3.1-8B-S2S-Omni", "architectures": [ "LlavaS2SLlamaForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 128000, "ctc_decoder_config": "(2,4096,32,11008)", "ctc_loss_weight": 1.0, "ctc_upsample_factor": 25, "eos_token_id": [ 128001, 128008, 128009 ], "freeze_speech_projector": false, "hidden_act": "silu", "hidden_size": 4096, "image_aspect_ratio": "anyres", "image_crop_resolution": null, "image_grid_pinpoints": [ [ 336, 672 ], [ 336, 1008 ], [ 336, 1344 ], [ 336, 1680 ], [ 336, 2016 ], [ 336, 2352 ], [ 336, 2688 ], [ 336, 3024 ], [ 336, 3360 ], [ 336, 3696 ], [ 336, 4032 ], [ 336, 4368 ], [ 336, 4704 ], [ 336, 5040 ], [ 336, 5376 ], [ 336, 5712 ], [ 336, 6048 ], [ 336, 6384 ], [ 336, 6720 ], [ 336, 7056 ], [ 336, 7392 ], [ 336, 7728 ], [ 336, 8064 ], [ 336, 8400 ], [ 336, 8736 ], [ 336, 9072 ], [ 336, 9408 ], [ 336, 9744 ], [ 336, 10080 ], [ 336, 10416 ], [ 336, 10752 ], [ 336, 11088 ], [ 336, 11424 ], [ 336, 11760 ], [ 336, 12096 ], [ 336, 12432 ], [ 336, 12768 ], [ 336, 13104 ], [ 336, 13440 ], [ 336, 13776 ], [ 336, 14112 ], [ 336, 14448 ], [ 336, 14784 ], [ 336, 15120 ], [ 336, 15456 ], [ 336, 15792 ], [ 336, 16128 ], [ 336, 16464 ], [ 672, 336 ], [ 672, 672 ], [ 672, 1008 ], [ 672, 1344 ], [ 672, 1680 ], [ 672, 2016 ], [ 672, 2352 ], [ 672, 2688 ], [ 672, 3024 ], [ 672, 3360 ], [ 672, 3696 ], [ 672, 4032 ], [ 672, 4368 ], [ 672, 4704 ], [ 672, 5040 ], [ 672, 5376 ], [ 672, 5712 ], [ 672, 6048 ], [ 672, 6384 ], [ 672, 6720 ], [ 672, 7056 ], [ 672, 7392 ], [ 672, 7728 ], [ 672, 8064 ], [ 1008, 336 ], [ 1008, 672 ], [ 1008, 1008 ], [ 1008, 1344 ], [ 1008, 1680 ], [ 1008, 2016 ], [ 1008, 2352 ], [ 1008, 2688 ], [ 1008, 3024 ], [ 1008, 3360 ], [ 1008, 3696 ], [ 1008, 4032 ], [ 1008, 4368 ], [ 1008, 4704 ], [ 1008, 5040 ], [ 1008, 5376 ], [ 1344, 336 ], [ 1344, 672 ], [ 1344, 1008 ], [ 1344, 1344 ], [ 1344, 1680 ], [ 1344, 2016 ], [ 1344, 2352 ], [ 1344, 2688 ], [ 1344, 3024 ], [ 1344, 3360 ], [ 1344, 3696 ], [ 1344, 4032 ], [ 1680, 336 ], [ 1680, 672 ], [ 1680, 1008 ], [ 1680, 1344 ], [ 1680, 1680 ], [ 1680, 2016 ], [ 1680, 2352 ], [ 1680, 2688 ], [ 1680, 3024 ], [ 2016, 336 ], [ 2016, 672 ], [ 2016, 1008 ], [ 2016, 1344 ], [ 2016, 1680 ], [ 2016, 2016 ], [ 2016, 2352 ], [ 2016, 2688 ], [ 2352, 336 ], [ 2352, 672 ], [ 2352, 1008 ], [ 2352, 1344 ], [ 2352, 1680 ], [ 2352, 2016 ], [ 2352, 2352 ], [ 2688, 336 ], [ 2688, 672 ], [ 2688, 1008 ], [ 2688, 1344 ], [ 2688, 1680 ], [ 2688, 2016 ], [ 3024, 336 ], [ 3024, 672 ], [ 3024, 1008 ], [ 3024, 1344 ], [ 3024, 1680 ], [ 3360, 336 ], [ 3360, 672 ], [ 3360, 1008 ], [ 3360, 1344 ], [ 3696, 336 ], [ 3696, 672 ], [ 3696, 1008 ], [ 3696, 1344 ], [ 4032, 336 ], [ 4032, 672 ], [ 4032, 1008 ], [ 4032, 1344 ], [ 4368, 336 ], [ 4368, 672 ], [ 4368, 1008 ], [ 4704, 336 ], [ 4704, 672 ], [ 4704, 1008 ], [ 5040, 336 ], [ 5040, 672 ], [ 5040, 1008 ], [ 5376, 336 ], [ 5376, 672 ], [ 5376, 1008 ], [ 5712, 336 ], [ 5712, 672 ], [ 6048, 336 ], [ 6048, 672 ], [ 6384, 336 ], [ 6384, 672 ], [ 6720, 336 ], [ 6720, 672 ], [ 7056, 336 ], [ 7056, 672 ], [ 7392, 336 ], [ 7392, 672 ], [ 7728, 336 ], [ 7728, 672 ], [ 8064, 336 ], [ 8064, 672 ], [ 8400, 336 ], [ 8736, 336 ], [ 9072, 336 ], [ 9408, 336 ], [ 9744, 336 ], [ 10080, 336 ], [ 10416, 336 ], [ 10752, 336 ], [ 11088, 336 ], [ 11424, 336 ], [ 11760, 336 ], [ 12096, 336 ], [ 12432, 336 ], [ 12768, 336 ], [ 13104, 336 ], [ 13440, 336 ], [ 13776, 336 ], [ 14112, 336 ], [ 14448, 336 ], [ 14784, 336 ], [ 15120, 336 ], [ 15456, 336 ], [ 15792, 336 ], [ 16128, 336 ], [ 16464, 336 ] ], "image_split_resolution": null, "initializer_range": 0.02, "intermediate_size": 14336, "max_position_embeddings": 131072, "mlp_bias": false, "mm_hidden_size": 1024, "mm_patch_merge_type": "unires", "mm_projector_lr": null, "mm_projector_type": "mlp2x_gelu", "mm_resampler_type": null, "mm_tunable_parts": "mm_mlp_adapter,mm_language_model", "mm_use_im_patch_token": false, "mm_use_im_start_end": false, "mm_vision_select_feature": "patch", "mm_vision_select_layer": -2, "mm_vision_tower": "checkpoints/clip-vit-large-patch14-336", "mm_vision_tower_lr": 2e-06, "model_type": "llama", "num_attention_heads": 32, "num_hidden_layers": 32, "num_key_value_heads": 8, "pos_skipping_range": 4096, "pretraining_tp": 1, "rms_norm_eps": 1e-05, "rope_scaling": { "factor": 8.0, "high_freq_factor": 4.0, "low_freq_factor": 1.0, "original_max_position_embeddings": 8192, "rope_type": "llama3" }, "rope_theta": 500000.0, "speech_encoder": "checkpoints/whisper/large-v3.pt", "speech_encoder_ds_rate": 5, "speech_encoder_hidden_size": 1280, "speech_encoder_type": "whisper", "speech_generator": null, "speech_generator_type": "ctc", "speech_normalize": false, "speech_projector_lr": null, "speech_projector_type": "linear", "tie_word_embeddings": false, "tokenizer_model_max_length": 16384, "tokenizer_padding_side": "right", "torch_dtype": "bfloat16", "transformers_version": "4.44.0", "tune_speech_generator_only": false, "tune_speech_projector": false, "unit_vocab_size": 1000, "use_cache": false, "use_mm_proj": true, "use_pos_skipping": false, "vision_tower_pretrained": null, "vocab_size": 128256 }