|
{ |
|
"add_cross_attention": false, |
|
"architectures": [ |
|
"Mistral3ForConditionalGeneration" |
|
], |
|
"bad_words_ids": null, |
|
"begin_suppress_tokens": null, |
|
"bos_token_id": null, |
|
"chunk_size_feed_forward": 0, |
|
"cross_attention_hidden_size": null, |
|
"decoder_start_token_id": null, |
|
"diversity_penalty": 0.0, |
|
"do_sample": false, |
|
"early_stopping": false, |
|
"encoder_no_repeat_ngram_size": 0, |
|
"eos_token_id": null, |
|
"exponential_decay_length_penalty": null, |
|
"finetuning_task": null, |
|
"forced_bos_token_id": null, |
|
"forced_eos_token_id": null, |
|
"id2label": { |
|
"0": "LABEL_0", |
|
"1": "LABEL_1" |
|
}, |
|
"image_token_index": 10, |
|
"is_decoder": false, |
|
"is_encoder_decoder": false, |
|
"label2id": { |
|
"LABEL_0": 0, |
|
"LABEL_1": 1 |
|
}, |
|
"length_penalty": 1.0, |
|
"max_length": 20, |
|
"min_length": 0, |
|
"model_type": "mistral3", |
|
"multimodal_projector_bias": false, |
|
"no_repeat_ngram_size": 0, |
|
"num_beam_groups": 1, |
|
"num_beams": 1, |
|
"num_return_sequences": 1, |
|
"output_attentions": false, |
|
"output_hidden_states": false, |
|
"output_scores": false, |
|
"pad_token_id": null, |
|
"prefix": null, |
|
"problem_type": null, |
|
"projector_hidden_act": "gelu", |
|
"pruned_heads": {}, |
|
"quantization": { |
|
"group_size": 64, |
|
"bits": 8, |
|
"vision_tower.vision_model.patch_conv": false, |
|
"vision_tower.vision_model.ln_pre": false, |
|
"vision_tower.vision_model.transformer.layers.0.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.0.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.0.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.0.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.0.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.0.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.0.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.0.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.0.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.1.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.1.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.1.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.1.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.1.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.1.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.1.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.1.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.1.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.2.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.2.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.2.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.2.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.2.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.2.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.2.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.2.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.2.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.3.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.3.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.3.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.3.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.3.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.3.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.3.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.3.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.3.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.4.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.4.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.4.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.4.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.4.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.4.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.4.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.4.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.4.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.5.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.5.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.5.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.5.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.5.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.5.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.5.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.5.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.5.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.6.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.6.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.6.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.6.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.6.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.6.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.6.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.6.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.6.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.7.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.7.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.7.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.7.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.7.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.7.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.7.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.7.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.7.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.8.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.8.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.8.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.8.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.8.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.8.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.8.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.8.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.8.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.9.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.9.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.9.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.9.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.9.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.9.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.9.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.9.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.9.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.10.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.10.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.10.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.10.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.10.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.10.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.10.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.10.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.10.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.11.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.11.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.11.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.11.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.11.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.11.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.11.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.11.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.11.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.12.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.12.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.12.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.12.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.12.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.12.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.12.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.12.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.12.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.13.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.13.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.13.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.13.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.13.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.13.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.13.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.13.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.13.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.14.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.14.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.14.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.14.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.14.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.14.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.14.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.14.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.14.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.15.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.15.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.15.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.15.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.15.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.15.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.15.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.15.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.15.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.16.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.16.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.16.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.16.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.16.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.16.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.16.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.16.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.16.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.17.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.17.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.17.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.17.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.17.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.17.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.17.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.17.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.17.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.18.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.18.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.18.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.18.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.18.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.18.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.18.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.18.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.18.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.19.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.19.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.19.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.19.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.19.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.19.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.19.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.19.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.19.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.20.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.20.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.20.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.20.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.20.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.20.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.20.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.20.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.20.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.21.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.21.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.21.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.21.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.21.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.21.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.21.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.21.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.21.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.22.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.22.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.22.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.22.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.22.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.22.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.22.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.22.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.22.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.23.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.23.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.23.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.23.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.23.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.23.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.23.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.23.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.23.ffn_norm": false, |
|
"language_model.model.embed_tokens": true, |
|
"language_model.model.layers.0.self_attn.q_proj": true, |
|
"language_model.model.layers.0.self_attn.k_proj": true, |
|
"language_model.model.layers.0.self_attn.v_proj": true, |
|
"language_model.model.layers.0.self_attn.o_proj": true, |
|
"language_model.model.layers.0.self_attn.rope": false, |
|
"language_model.model.layers.0.mlp.gate_proj": true, |
|
"language_model.model.layers.0.mlp.down_proj": true, |
|
"language_model.model.layers.0.mlp.up_proj": true, |
|
"language_model.model.layers.0.input_layernorm": false, |
|
"language_model.model.layers.0.post_attention_layernorm": false, |
|
"language_model.model.layers.1.self_attn.q_proj": true, |
|
"language_model.model.layers.1.self_attn.k_proj": true, |
|
"language_model.model.layers.1.self_attn.v_proj": true, |
|
"language_model.model.layers.1.self_attn.o_proj": true, |
|
"language_model.model.layers.1.self_attn.rope": false, |
|
"language_model.model.layers.1.mlp.gate_proj": true, |
|
"language_model.model.layers.1.mlp.down_proj": true, |
|
"language_model.model.layers.1.mlp.up_proj": true, |
|
"language_model.model.layers.1.input_layernorm": false, |
|
"language_model.model.layers.1.post_attention_layernorm": false, |
|
"language_model.model.layers.2.self_attn.q_proj": true, |
|
"language_model.model.layers.2.self_attn.k_proj": true, |
|
"language_model.model.layers.2.self_attn.v_proj": true, |
|
"language_model.model.layers.2.self_attn.o_proj": true, |
|
"language_model.model.layers.2.self_attn.rope": false, |
|
"language_model.model.layers.2.mlp.gate_proj": true, |
|
"language_model.model.layers.2.mlp.down_proj": true, |
|
"language_model.model.layers.2.mlp.up_proj": true, |
|
"language_model.model.layers.2.input_layernorm": false, |
|
"language_model.model.layers.2.post_attention_layernorm": false, |
|
"language_model.model.layers.3.self_attn.q_proj": true, |
|
"language_model.model.layers.3.self_attn.k_proj": true, |
|
"language_model.model.layers.3.self_attn.v_proj": true, |
|
"language_model.model.layers.3.self_attn.o_proj": true, |
|
"language_model.model.layers.3.self_attn.rope": false, |
|
"language_model.model.layers.3.mlp.gate_proj": true, |
|
"language_model.model.layers.3.mlp.down_proj": true, |
|
"language_model.model.layers.3.mlp.up_proj": true, |
|
"language_model.model.layers.3.input_layernorm": false, |
|
"language_model.model.layers.3.post_attention_layernorm": false, |
|
"language_model.model.layers.4.self_attn.q_proj": true, |
|
"language_model.model.layers.4.self_attn.k_proj": true, |
|
"language_model.model.layers.4.self_attn.v_proj": true, |
|
"language_model.model.layers.4.self_attn.o_proj": true, |
|
"language_model.model.layers.4.self_attn.rope": false, |
|
"language_model.model.layers.4.mlp.gate_proj": true, |
|
"language_model.model.layers.4.mlp.down_proj": true, |
|
"language_model.model.layers.4.mlp.up_proj": true, |
|
"language_model.model.layers.4.input_layernorm": false, |
|
"language_model.model.layers.4.post_attention_layernorm": false, |
|
"language_model.model.layers.5.self_attn.q_proj": true, |
|
"language_model.model.layers.5.self_attn.k_proj": true, |
|
"language_model.model.layers.5.self_attn.v_proj": true, |
|
"language_model.model.layers.5.self_attn.o_proj": true, |
|
"language_model.model.layers.5.self_attn.rope": false, |
|
"language_model.model.layers.5.mlp.gate_proj": true, |
|
"language_model.model.layers.5.mlp.down_proj": true, |
|
"language_model.model.layers.5.mlp.up_proj": true, |
|
"language_model.model.layers.5.input_layernorm": false, |
|
"language_model.model.layers.5.post_attention_layernorm": false, |
|
"language_model.model.layers.6.self_attn.q_proj": true, |
|
"language_model.model.layers.6.self_attn.k_proj": true, |
|
"language_model.model.layers.6.self_attn.v_proj": true, |
|
"language_model.model.layers.6.self_attn.o_proj": true, |
|
"language_model.model.layers.6.self_attn.rope": false, |
|
"language_model.model.layers.6.mlp.gate_proj": true, |
|
"language_model.model.layers.6.mlp.down_proj": true, |
|
"language_model.model.layers.6.mlp.up_proj": true, |
|
"language_model.model.layers.6.input_layernorm": false, |
|
"language_model.model.layers.6.post_attention_layernorm": false, |
|
"language_model.model.layers.7.self_attn.q_proj": true, |
|
"language_model.model.layers.7.self_attn.k_proj": true, |
|
"language_model.model.layers.7.self_attn.v_proj": true, |
|
"language_model.model.layers.7.self_attn.o_proj": true, |
|
"language_model.model.layers.7.self_attn.rope": false, |
|
"language_model.model.layers.7.mlp.gate_proj": true, |
|
"language_model.model.layers.7.mlp.down_proj": true, |
|
"language_model.model.layers.7.mlp.up_proj": true, |
|
"language_model.model.layers.7.input_layernorm": false, |
|
"language_model.model.layers.7.post_attention_layernorm": false, |
|
"language_model.model.layers.8.self_attn.q_proj": true, |
|
"language_model.model.layers.8.self_attn.k_proj": true, |
|
"language_model.model.layers.8.self_attn.v_proj": true, |
|
"language_model.model.layers.8.self_attn.o_proj": true, |
|
"language_model.model.layers.8.self_attn.rope": false, |
|
"language_model.model.layers.8.mlp.gate_proj": true, |
|
"language_model.model.layers.8.mlp.down_proj": true, |
|
"language_model.model.layers.8.mlp.up_proj": true, |
|
"language_model.model.layers.8.input_layernorm": false, |
|
"language_model.model.layers.8.post_attention_layernorm": false, |
|
"language_model.model.layers.9.self_attn.q_proj": true, |
|
"language_model.model.layers.9.self_attn.k_proj": true, |
|
"language_model.model.layers.9.self_attn.v_proj": true, |
|
"language_model.model.layers.9.self_attn.o_proj": true, |
|
"language_model.model.layers.9.self_attn.rope": false, |
|
"language_model.model.layers.9.mlp.gate_proj": true, |
|
"language_model.model.layers.9.mlp.down_proj": true, |
|
"language_model.model.layers.9.mlp.up_proj": true, |
|
"language_model.model.layers.9.input_layernorm": false, |
|
"language_model.model.layers.9.post_attention_layernorm": false, |
|
"language_model.model.layers.10.self_attn.q_proj": true, |
|
"language_model.model.layers.10.self_attn.k_proj": true, |
|
"language_model.model.layers.10.self_attn.v_proj": true, |
|
"language_model.model.layers.10.self_attn.o_proj": true, |
|
"language_model.model.layers.10.self_attn.rope": false, |
|
"language_model.model.layers.10.mlp.gate_proj": true, |
|
"language_model.model.layers.10.mlp.down_proj": true, |
|
"language_model.model.layers.10.mlp.up_proj": true, |
|
"language_model.model.layers.10.input_layernorm": false, |
|
"language_model.model.layers.10.post_attention_layernorm": false, |
|
"language_model.model.layers.11.self_attn.q_proj": true, |
|
"language_model.model.layers.11.self_attn.k_proj": true, |
|
"language_model.model.layers.11.self_attn.v_proj": true, |
|
"language_model.model.layers.11.self_attn.o_proj": true, |
|
"language_model.model.layers.11.self_attn.rope": false, |
|
"language_model.model.layers.11.mlp.gate_proj": true, |
|
"language_model.model.layers.11.mlp.down_proj": true, |
|
"language_model.model.layers.11.mlp.up_proj": true, |
|
"language_model.model.layers.11.input_layernorm": false, |
|
"language_model.model.layers.11.post_attention_layernorm": false, |
|
"language_model.model.layers.12.self_attn.q_proj": true, |
|
"language_model.model.layers.12.self_attn.k_proj": true, |
|
"language_model.model.layers.12.self_attn.v_proj": true, |
|
"language_model.model.layers.12.self_attn.o_proj": true, |
|
"language_model.model.layers.12.self_attn.rope": false, |
|
"language_model.model.layers.12.mlp.gate_proj": true, |
|
"language_model.model.layers.12.mlp.down_proj": true, |
|
"language_model.model.layers.12.mlp.up_proj": true, |
|
"language_model.model.layers.12.input_layernorm": false, |
|
"language_model.model.layers.12.post_attention_layernorm": false, |
|
"language_model.model.layers.13.self_attn.q_proj": true, |
|
"language_model.model.layers.13.self_attn.k_proj": true, |
|
"language_model.model.layers.13.self_attn.v_proj": true, |
|
"language_model.model.layers.13.self_attn.o_proj": true, |
|
"language_model.model.layers.13.self_attn.rope": false, |
|
"language_model.model.layers.13.mlp.gate_proj": true, |
|
"language_model.model.layers.13.mlp.down_proj": true, |
|
"language_model.model.layers.13.mlp.up_proj": true, |
|
"language_model.model.layers.13.input_layernorm": false, |
|
"language_model.model.layers.13.post_attention_layernorm": false, |
|
"language_model.model.layers.14.self_attn.q_proj": true, |
|
"language_model.model.layers.14.self_attn.k_proj": true, |
|
"language_model.model.layers.14.self_attn.v_proj": true, |
|
"language_model.model.layers.14.self_attn.o_proj": true, |
|
"language_model.model.layers.14.self_attn.rope": false, |
|
"language_model.model.layers.14.mlp.gate_proj": true, |
|
"language_model.model.layers.14.mlp.down_proj": true, |
|
"language_model.model.layers.14.mlp.up_proj": true, |
|
"language_model.model.layers.14.input_layernorm": false, |
|
"language_model.model.layers.14.post_attention_layernorm": false, |
|
"language_model.model.layers.15.self_attn.q_proj": true, |
|
"language_model.model.layers.15.self_attn.k_proj": true, |
|
"language_model.model.layers.15.self_attn.v_proj": true, |
|
"language_model.model.layers.15.self_attn.o_proj": true, |
|
"language_model.model.layers.15.self_attn.rope": false, |
|
"language_model.model.layers.15.mlp.gate_proj": true, |
|
"language_model.model.layers.15.mlp.down_proj": true, |
|
"language_model.model.layers.15.mlp.up_proj": true, |
|
"language_model.model.layers.15.input_layernorm": false, |
|
"language_model.model.layers.15.post_attention_layernorm": false, |
|
"language_model.model.layers.16.self_attn.q_proj": true, |
|
"language_model.model.layers.16.self_attn.k_proj": true, |
|
"language_model.model.layers.16.self_attn.v_proj": true, |
|
"language_model.model.layers.16.self_attn.o_proj": true, |
|
"language_model.model.layers.16.self_attn.rope": false, |
|
"language_model.model.layers.16.mlp.gate_proj": true, |
|
"language_model.model.layers.16.mlp.down_proj": true, |
|
"language_model.model.layers.16.mlp.up_proj": true, |
|
"language_model.model.layers.16.input_layernorm": false, |
|
"language_model.model.layers.16.post_attention_layernorm": false, |
|
"language_model.model.layers.17.self_attn.q_proj": true, |
|
"language_model.model.layers.17.self_attn.k_proj": true, |
|
"language_model.model.layers.17.self_attn.v_proj": true, |
|
"language_model.model.layers.17.self_attn.o_proj": true, |
|
"language_model.model.layers.17.self_attn.rope": false, |
|
"language_model.model.layers.17.mlp.gate_proj": true, |
|
"language_model.model.layers.17.mlp.down_proj": true, |
|
"language_model.model.layers.17.mlp.up_proj": true, |
|
"language_model.model.layers.17.input_layernorm": false, |
|
"language_model.model.layers.17.post_attention_layernorm": false, |
|
"language_model.model.layers.18.self_attn.q_proj": true, |
|
"language_model.model.layers.18.self_attn.k_proj": true, |
|
"language_model.model.layers.18.self_attn.v_proj": true, |
|
"language_model.model.layers.18.self_attn.o_proj": true, |
|
"language_model.model.layers.18.self_attn.rope": false, |
|
"language_model.model.layers.18.mlp.gate_proj": true, |
|
"language_model.model.layers.18.mlp.down_proj": true, |
|
"language_model.model.layers.18.mlp.up_proj": true, |
|
"language_model.model.layers.18.input_layernorm": false, |
|
"language_model.model.layers.18.post_attention_layernorm": false, |
|
"language_model.model.layers.19.self_attn.q_proj": true, |
|
"language_model.model.layers.19.self_attn.k_proj": true, |
|
"language_model.model.layers.19.self_attn.v_proj": true, |
|
"language_model.model.layers.19.self_attn.o_proj": true, |
|
"language_model.model.layers.19.self_attn.rope": false, |
|
"language_model.model.layers.19.mlp.gate_proj": true, |
|
"language_model.model.layers.19.mlp.down_proj": true, |
|
"language_model.model.layers.19.mlp.up_proj": true, |
|
"language_model.model.layers.19.input_layernorm": false, |
|
"language_model.model.layers.19.post_attention_layernorm": false, |
|
"language_model.model.layers.20.self_attn.q_proj": true, |
|
"language_model.model.layers.20.self_attn.k_proj": true, |
|
"language_model.model.layers.20.self_attn.v_proj": true, |
|
"language_model.model.layers.20.self_attn.o_proj": true, |
|
"language_model.model.layers.20.self_attn.rope": false, |
|
"language_model.model.layers.20.mlp.gate_proj": true, |
|
"language_model.model.layers.20.mlp.down_proj": true, |
|
"language_model.model.layers.20.mlp.up_proj": true, |
|
"language_model.model.layers.20.input_layernorm": false, |
|
"language_model.model.layers.20.post_attention_layernorm": false, |
|
"language_model.model.layers.21.self_attn.q_proj": true, |
|
"language_model.model.layers.21.self_attn.k_proj": true, |
|
"language_model.model.layers.21.self_attn.v_proj": true, |
|
"language_model.model.layers.21.self_attn.o_proj": true, |
|
"language_model.model.layers.21.self_attn.rope": false, |
|
"language_model.model.layers.21.mlp.gate_proj": true, |
|
"language_model.model.layers.21.mlp.down_proj": true, |
|
"language_model.model.layers.21.mlp.up_proj": true, |
|
"language_model.model.layers.21.input_layernorm": false, |
|
"language_model.model.layers.21.post_attention_layernorm": false, |
|
"language_model.model.layers.22.self_attn.q_proj": true, |
|
"language_model.model.layers.22.self_attn.k_proj": true, |
|
"language_model.model.layers.22.self_attn.v_proj": true, |
|
"language_model.model.layers.22.self_attn.o_proj": true, |
|
"language_model.model.layers.22.self_attn.rope": false, |
|
"language_model.model.layers.22.mlp.gate_proj": true, |
|
"language_model.model.layers.22.mlp.down_proj": true, |
|
"language_model.model.layers.22.mlp.up_proj": true, |
|
"language_model.model.layers.22.input_layernorm": false, |
|
"language_model.model.layers.22.post_attention_layernorm": false, |
|
"language_model.model.layers.23.self_attn.q_proj": true, |
|
"language_model.model.layers.23.self_attn.k_proj": true, |
|
"language_model.model.layers.23.self_attn.v_proj": true, |
|
"language_model.model.layers.23.self_attn.o_proj": true, |
|
"language_model.model.layers.23.self_attn.rope": false, |
|
"language_model.model.layers.23.mlp.gate_proj": true, |
|
"language_model.model.layers.23.mlp.down_proj": true, |
|
"language_model.model.layers.23.mlp.up_proj": true, |
|
"language_model.model.layers.23.input_layernorm": false, |
|
"language_model.model.layers.23.post_attention_layernorm": false, |
|
"language_model.model.layers.24.self_attn.q_proj": true, |
|
"language_model.model.layers.24.self_attn.k_proj": true, |
|
"language_model.model.layers.24.self_attn.v_proj": true, |
|
"language_model.model.layers.24.self_attn.o_proj": true, |
|
"language_model.model.layers.24.self_attn.rope": false, |
|
"language_model.model.layers.24.mlp.gate_proj": true, |
|
"language_model.model.layers.24.mlp.down_proj": true, |
|
"language_model.model.layers.24.mlp.up_proj": true, |
|
"language_model.model.layers.24.input_layernorm": false, |
|
"language_model.model.layers.24.post_attention_layernorm": false, |
|
"language_model.model.layers.25.self_attn.q_proj": true, |
|
"language_model.model.layers.25.self_attn.k_proj": true, |
|
"language_model.model.layers.25.self_attn.v_proj": true, |
|
"language_model.model.layers.25.self_attn.o_proj": true, |
|
"language_model.model.layers.25.self_attn.rope": false, |
|
"language_model.model.layers.25.mlp.gate_proj": true, |
|
"language_model.model.layers.25.mlp.down_proj": true, |
|
"language_model.model.layers.25.mlp.up_proj": true, |
|
"language_model.model.layers.25.input_layernorm": false, |
|
"language_model.model.layers.25.post_attention_layernorm": false, |
|
"language_model.model.layers.26.self_attn.q_proj": true, |
|
"language_model.model.layers.26.self_attn.k_proj": true, |
|
"language_model.model.layers.26.self_attn.v_proj": true, |
|
"language_model.model.layers.26.self_attn.o_proj": true, |
|
"language_model.model.layers.26.self_attn.rope": false, |
|
"language_model.model.layers.26.mlp.gate_proj": true, |
|
"language_model.model.layers.26.mlp.down_proj": true, |
|
"language_model.model.layers.26.mlp.up_proj": true, |
|
"language_model.model.layers.26.input_layernorm": false, |
|
"language_model.model.layers.26.post_attention_layernorm": false, |
|
"language_model.model.layers.27.self_attn.q_proj": true, |
|
"language_model.model.layers.27.self_attn.k_proj": true, |
|
"language_model.model.layers.27.self_attn.v_proj": true, |
|
"language_model.model.layers.27.self_attn.o_proj": true, |
|
"language_model.model.layers.27.self_attn.rope": false, |
|
"language_model.model.layers.27.mlp.gate_proj": true, |
|
"language_model.model.layers.27.mlp.down_proj": true, |
|
"language_model.model.layers.27.mlp.up_proj": true, |
|
"language_model.model.layers.27.input_layernorm": false, |
|
"language_model.model.layers.27.post_attention_layernorm": false, |
|
"language_model.model.layers.28.self_attn.q_proj": true, |
|
"language_model.model.layers.28.self_attn.k_proj": true, |
|
"language_model.model.layers.28.self_attn.v_proj": true, |
|
"language_model.model.layers.28.self_attn.o_proj": true, |
|
"language_model.model.layers.28.self_attn.rope": false, |
|
"language_model.model.layers.28.mlp.gate_proj": true, |
|
"language_model.model.layers.28.mlp.down_proj": true, |
|
"language_model.model.layers.28.mlp.up_proj": true, |
|
"language_model.model.layers.28.input_layernorm": false, |
|
"language_model.model.layers.28.post_attention_layernorm": false, |
|
"language_model.model.layers.29.self_attn.q_proj": true, |
|
"language_model.model.layers.29.self_attn.k_proj": true, |
|
"language_model.model.layers.29.self_attn.v_proj": true, |
|
"language_model.model.layers.29.self_attn.o_proj": true, |
|
"language_model.model.layers.29.self_attn.rope": false, |
|
"language_model.model.layers.29.mlp.gate_proj": true, |
|
"language_model.model.layers.29.mlp.down_proj": true, |
|
"language_model.model.layers.29.mlp.up_proj": true, |
|
"language_model.model.layers.29.input_layernorm": false, |
|
"language_model.model.layers.29.post_attention_layernorm": false, |
|
"language_model.model.layers.30.self_attn.q_proj": true, |
|
"language_model.model.layers.30.self_attn.k_proj": true, |
|
"language_model.model.layers.30.self_attn.v_proj": true, |
|
"language_model.model.layers.30.self_attn.o_proj": true, |
|
"language_model.model.layers.30.self_attn.rope": false, |
|
"language_model.model.layers.30.mlp.gate_proj": true, |
|
"language_model.model.layers.30.mlp.down_proj": true, |
|
"language_model.model.layers.30.mlp.up_proj": true, |
|
"language_model.model.layers.30.input_layernorm": false, |
|
"language_model.model.layers.30.post_attention_layernorm": false, |
|
"language_model.model.layers.31.self_attn.q_proj": true, |
|
"language_model.model.layers.31.self_attn.k_proj": true, |
|
"language_model.model.layers.31.self_attn.v_proj": true, |
|
"language_model.model.layers.31.self_attn.o_proj": true, |
|
"language_model.model.layers.31.self_attn.rope": false, |
|
"language_model.model.layers.31.mlp.gate_proj": true, |
|
"language_model.model.layers.31.mlp.down_proj": true, |
|
"language_model.model.layers.31.mlp.up_proj": true, |
|
"language_model.model.layers.31.input_layernorm": false, |
|
"language_model.model.layers.31.post_attention_layernorm": false, |
|
"language_model.model.layers.32.self_attn.q_proj": true, |
|
"language_model.model.layers.32.self_attn.k_proj": true, |
|
"language_model.model.layers.32.self_attn.v_proj": true, |
|
"language_model.model.layers.32.self_attn.o_proj": true, |
|
"language_model.model.layers.32.self_attn.rope": false, |
|
"language_model.model.layers.32.mlp.gate_proj": true, |
|
"language_model.model.layers.32.mlp.down_proj": true, |
|
"language_model.model.layers.32.mlp.up_proj": true, |
|
"language_model.model.layers.32.input_layernorm": false, |
|
"language_model.model.layers.32.post_attention_layernorm": false, |
|
"language_model.model.layers.33.self_attn.q_proj": true, |
|
"language_model.model.layers.33.self_attn.k_proj": true, |
|
"language_model.model.layers.33.self_attn.v_proj": true, |
|
"language_model.model.layers.33.self_attn.o_proj": true, |
|
"language_model.model.layers.33.self_attn.rope": false, |
|
"language_model.model.layers.33.mlp.gate_proj": true, |
|
"language_model.model.layers.33.mlp.down_proj": true, |
|
"language_model.model.layers.33.mlp.up_proj": true, |
|
"language_model.model.layers.33.input_layernorm": false, |
|
"language_model.model.layers.33.post_attention_layernorm": false, |
|
"language_model.model.layers.34.self_attn.q_proj": true, |
|
"language_model.model.layers.34.self_attn.k_proj": true, |
|
"language_model.model.layers.34.self_attn.v_proj": true, |
|
"language_model.model.layers.34.self_attn.o_proj": true, |
|
"language_model.model.layers.34.self_attn.rope": false, |
|
"language_model.model.layers.34.mlp.gate_proj": true, |
|
"language_model.model.layers.34.mlp.down_proj": true, |
|
"language_model.model.layers.34.mlp.up_proj": true, |
|
"language_model.model.layers.34.input_layernorm": false, |
|
"language_model.model.layers.34.post_attention_layernorm": false, |
|
"language_model.model.layers.35.self_attn.q_proj": true, |
|
"language_model.model.layers.35.self_attn.k_proj": true, |
|
"language_model.model.layers.35.self_attn.v_proj": true, |
|
"language_model.model.layers.35.self_attn.o_proj": true, |
|
"language_model.model.layers.35.self_attn.rope": false, |
|
"language_model.model.layers.35.mlp.gate_proj": true, |
|
"language_model.model.layers.35.mlp.down_proj": true, |
|
"language_model.model.layers.35.mlp.up_proj": true, |
|
"language_model.model.layers.35.input_layernorm": false, |
|
"language_model.model.layers.35.post_attention_layernorm": false, |
|
"language_model.model.layers.36.self_attn.q_proj": true, |
|
"language_model.model.layers.36.self_attn.k_proj": true, |
|
"language_model.model.layers.36.self_attn.v_proj": true, |
|
"language_model.model.layers.36.self_attn.o_proj": true, |
|
"language_model.model.layers.36.self_attn.rope": false, |
|
"language_model.model.layers.36.mlp.gate_proj": true, |
|
"language_model.model.layers.36.mlp.down_proj": true, |
|
"language_model.model.layers.36.mlp.up_proj": true, |
|
"language_model.model.layers.36.input_layernorm": false, |
|
"language_model.model.layers.36.post_attention_layernorm": false, |
|
"language_model.model.layers.37.self_attn.q_proj": true, |
|
"language_model.model.layers.37.self_attn.k_proj": true, |
|
"language_model.model.layers.37.self_attn.v_proj": true, |
|
"language_model.model.layers.37.self_attn.o_proj": true, |
|
"language_model.model.layers.37.self_attn.rope": false, |
|
"language_model.model.layers.37.mlp.gate_proj": true, |
|
"language_model.model.layers.37.mlp.down_proj": true, |
|
"language_model.model.layers.37.mlp.up_proj": true, |
|
"language_model.model.layers.37.input_layernorm": false, |
|
"language_model.model.layers.37.post_attention_layernorm": false, |
|
"language_model.model.layers.38.self_attn.q_proj": true, |
|
"language_model.model.layers.38.self_attn.k_proj": true, |
|
"language_model.model.layers.38.self_attn.v_proj": true, |
|
"language_model.model.layers.38.self_attn.o_proj": true, |
|
"language_model.model.layers.38.self_attn.rope": false, |
|
"language_model.model.layers.38.mlp.gate_proj": true, |
|
"language_model.model.layers.38.mlp.down_proj": true, |
|
"language_model.model.layers.38.mlp.up_proj": true, |
|
"language_model.model.layers.38.input_layernorm": false, |
|
"language_model.model.layers.38.post_attention_layernorm": false, |
|
"language_model.model.layers.39.self_attn.q_proj": true, |
|
"language_model.model.layers.39.self_attn.k_proj": true, |
|
"language_model.model.layers.39.self_attn.v_proj": true, |
|
"language_model.model.layers.39.self_attn.o_proj": true, |
|
"language_model.model.layers.39.self_attn.rope": false, |
|
"language_model.model.layers.39.mlp.gate_proj": true, |
|
"language_model.model.layers.39.mlp.down_proj": true, |
|
"language_model.model.layers.39.mlp.up_proj": true, |
|
"language_model.model.layers.39.input_layernorm": false, |
|
"language_model.model.layers.39.post_attention_layernorm": false, |
|
"language_model.model.norm": false, |
|
"language_model.lm_head": true, |
|
"multi_modal_projector.norm": false, |
|
"multi_modal_projector.patch_merger.merging_layer": true, |
|
"multi_modal_projector.linear_1": true, |
|
"multi_modal_projector.gelu": false, |
|
"multi_modal_projector.linear_2": true |
|
}, |
|
"quantization_config": { |
|
"group_size": 64, |
|
"bits": 8, |
|
"vision_tower.vision_model.patch_conv": false, |
|
"vision_tower.vision_model.ln_pre": false, |
|
"vision_tower.vision_model.transformer.layers.0.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.0.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.0.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.0.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.0.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.0.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.0.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.0.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.0.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.1.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.1.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.1.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.1.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.1.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.1.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.1.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.1.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.1.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.2.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.2.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.2.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.2.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.2.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.2.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.2.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.2.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.2.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.3.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.3.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.3.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.3.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.3.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.3.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.3.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.3.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.3.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.4.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.4.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.4.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.4.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.4.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.4.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.4.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.4.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.4.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.5.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.5.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.5.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.5.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.5.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.5.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.5.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.5.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.5.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.6.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.6.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.6.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.6.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.6.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.6.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.6.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.6.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.6.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.7.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.7.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.7.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.7.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.7.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.7.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.7.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.7.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.7.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.8.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.8.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.8.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.8.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.8.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.8.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.8.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.8.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.8.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.9.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.9.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.9.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.9.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.9.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.9.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.9.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.9.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.9.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.10.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.10.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.10.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.10.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.10.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.10.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.10.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.10.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.10.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.11.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.11.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.11.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.11.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.11.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.11.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.11.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.11.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.11.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.12.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.12.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.12.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.12.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.12.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.12.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.12.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.12.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.12.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.13.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.13.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.13.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.13.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.13.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.13.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.13.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.13.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.13.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.14.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.14.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.14.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.14.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.14.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.14.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.14.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.14.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.14.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.15.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.15.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.15.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.15.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.15.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.15.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.15.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.15.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.15.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.16.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.16.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.16.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.16.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.16.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.16.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.16.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.16.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.16.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.17.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.17.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.17.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.17.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.17.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.17.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.17.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.17.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.17.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.18.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.18.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.18.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.18.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.18.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.18.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.18.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.18.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.18.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.19.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.19.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.19.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.19.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.19.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.19.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.19.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.19.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.19.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.20.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.20.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.20.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.20.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.20.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.20.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.20.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.20.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.20.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.21.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.21.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.21.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.21.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.21.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.21.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.21.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.21.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.21.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.22.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.22.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.22.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.22.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.22.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.22.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.22.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.22.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.22.ffn_norm": false, |
|
"vision_tower.vision_model.transformer.layers.23.attention.k_proj": false, |
|
"vision_tower.vision_model.transformer.layers.23.attention.v_proj": false, |
|
"vision_tower.vision_model.transformer.layers.23.attention.q_proj": false, |
|
"vision_tower.vision_model.transformer.layers.23.attention.o_proj": false, |
|
"vision_tower.vision_model.transformer.layers.23.attention_norm": false, |
|
"vision_tower.vision_model.transformer.layers.23.feed_forward.gate_proj": false, |
|
"vision_tower.vision_model.transformer.layers.23.feed_forward.down_proj": false, |
|
"vision_tower.vision_model.transformer.layers.23.feed_forward.up_proj": false, |
|
"vision_tower.vision_model.transformer.layers.23.ffn_norm": false, |
|
"language_model.model.embed_tokens": true, |
|
"language_model.model.layers.0.self_attn.q_proj": true, |
|
"language_model.model.layers.0.self_attn.k_proj": true, |
|
"language_model.model.layers.0.self_attn.v_proj": true, |
|
"language_model.model.layers.0.self_attn.o_proj": true, |
|
"language_model.model.layers.0.self_attn.rope": false, |
|
"language_model.model.layers.0.mlp.gate_proj": true, |
|
"language_model.model.layers.0.mlp.down_proj": true, |
|
"language_model.model.layers.0.mlp.up_proj": true, |
|
"language_model.model.layers.0.input_layernorm": false, |
|
"language_model.model.layers.0.post_attention_layernorm": false, |
|
"language_model.model.layers.1.self_attn.q_proj": true, |
|
"language_model.model.layers.1.self_attn.k_proj": true, |
|
"language_model.model.layers.1.self_attn.v_proj": true, |
|
"language_model.model.layers.1.self_attn.o_proj": true, |
|
"language_model.model.layers.1.self_attn.rope": false, |
|
"language_model.model.layers.1.mlp.gate_proj": true, |
|
"language_model.model.layers.1.mlp.down_proj": true, |
|
"language_model.model.layers.1.mlp.up_proj": true, |
|
"language_model.model.layers.1.input_layernorm": false, |
|
"language_model.model.layers.1.post_attention_layernorm": false, |
|
"language_model.model.layers.2.self_attn.q_proj": true, |
|
"language_model.model.layers.2.self_attn.k_proj": true, |
|
"language_model.model.layers.2.self_attn.v_proj": true, |
|
"language_model.model.layers.2.self_attn.o_proj": true, |
|
"language_model.model.layers.2.self_attn.rope": false, |
|
"language_model.model.layers.2.mlp.gate_proj": true, |
|
"language_model.model.layers.2.mlp.down_proj": true, |
|
"language_model.model.layers.2.mlp.up_proj": true, |
|
"language_model.model.layers.2.input_layernorm": false, |
|
"language_model.model.layers.2.post_attention_layernorm": false, |
|
"language_model.model.layers.3.self_attn.q_proj": true, |
|
"language_model.model.layers.3.self_attn.k_proj": true, |
|
"language_model.model.layers.3.self_attn.v_proj": true, |
|
"language_model.model.layers.3.self_attn.o_proj": true, |
|
"language_model.model.layers.3.self_attn.rope": false, |
|
"language_model.model.layers.3.mlp.gate_proj": true, |
|
"language_model.model.layers.3.mlp.down_proj": true, |
|
"language_model.model.layers.3.mlp.up_proj": true, |
|
"language_model.model.layers.3.input_layernorm": false, |
|
"language_model.model.layers.3.post_attention_layernorm": false, |
|
"language_model.model.layers.4.self_attn.q_proj": true, |
|
"language_model.model.layers.4.self_attn.k_proj": true, |
|
"language_model.model.layers.4.self_attn.v_proj": true, |
|
"language_model.model.layers.4.self_attn.o_proj": true, |
|
"language_model.model.layers.4.self_attn.rope": false, |
|
"language_model.model.layers.4.mlp.gate_proj": true, |
|
"language_model.model.layers.4.mlp.down_proj": true, |
|
"language_model.model.layers.4.mlp.up_proj": true, |
|
"language_model.model.layers.4.input_layernorm": false, |
|
"language_model.model.layers.4.post_attention_layernorm": false, |
|
"language_model.model.layers.5.self_attn.q_proj": true, |
|
"language_model.model.layers.5.self_attn.k_proj": true, |
|
"language_model.model.layers.5.self_attn.v_proj": true, |
|
"language_model.model.layers.5.self_attn.o_proj": true, |
|
"language_model.model.layers.5.self_attn.rope": false, |
|
"language_model.model.layers.5.mlp.gate_proj": true, |
|
"language_model.model.layers.5.mlp.down_proj": true, |
|
"language_model.model.layers.5.mlp.up_proj": true, |
|
"language_model.model.layers.5.input_layernorm": false, |
|
"language_model.model.layers.5.post_attention_layernorm": false, |
|
"language_model.model.layers.6.self_attn.q_proj": true, |
|
"language_model.model.layers.6.self_attn.k_proj": true, |
|
"language_model.model.layers.6.self_attn.v_proj": true, |
|
"language_model.model.layers.6.self_attn.o_proj": true, |
|
"language_model.model.layers.6.self_attn.rope": false, |
|
"language_model.model.layers.6.mlp.gate_proj": true, |
|
"language_model.model.layers.6.mlp.down_proj": true, |
|
"language_model.model.layers.6.mlp.up_proj": true, |
|
"language_model.model.layers.6.input_layernorm": false, |
|
"language_model.model.layers.6.post_attention_layernorm": false, |
|
"language_model.model.layers.7.self_attn.q_proj": true, |
|
"language_model.model.layers.7.self_attn.k_proj": true, |
|
"language_model.model.layers.7.self_attn.v_proj": true, |
|
"language_model.model.layers.7.self_attn.o_proj": true, |
|
"language_model.model.layers.7.self_attn.rope": false, |
|
"language_model.model.layers.7.mlp.gate_proj": true, |
|
"language_model.model.layers.7.mlp.down_proj": true, |
|
"language_model.model.layers.7.mlp.up_proj": true, |
|
"language_model.model.layers.7.input_layernorm": false, |
|
"language_model.model.layers.7.post_attention_layernorm": false, |
|
"language_model.model.layers.8.self_attn.q_proj": true, |
|
"language_model.model.layers.8.self_attn.k_proj": true, |
|
"language_model.model.layers.8.self_attn.v_proj": true, |
|
"language_model.model.layers.8.self_attn.o_proj": true, |
|
"language_model.model.layers.8.self_attn.rope": false, |
|
"language_model.model.layers.8.mlp.gate_proj": true, |
|
"language_model.model.layers.8.mlp.down_proj": true, |
|
"language_model.model.layers.8.mlp.up_proj": true, |
|
"language_model.model.layers.8.input_layernorm": false, |
|
"language_model.model.layers.8.post_attention_layernorm": false, |
|
"language_model.model.layers.9.self_attn.q_proj": true, |
|
"language_model.model.layers.9.self_attn.k_proj": true, |
|
"language_model.model.layers.9.self_attn.v_proj": true, |
|
"language_model.model.layers.9.self_attn.o_proj": true, |
|
"language_model.model.layers.9.self_attn.rope": false, |
|
"language_model.model.layers.9.mlp.gate_proj": true, |
|
"language_model.model.layers.9.mlp.down_proj": true, |
|
"language_model.model.layers.9.mlp.up_proj": true, |
|
"language_model.model.layers.9.input_layernorm": false, |
|
"language_model.model.layers.9.post_attention_layernorm": false, |
|
"language_model.model.layers.10.self_attn.q_proj": true, |
|
"language_model.model.layers.10.self_attn.k_proj": true, |
|
"language_model.model.layers.10.self_attn.v_proj": true, |
|
"language_model.model.layers.10.self_attn.o_proj": true, |
|
"language_model.model.layers.10.self_attn.rope": false, |
|
"language_model.model.layers.10.mlp.gate_proj": true, |
|
"language_model.model.layers.10.mlp.down_proj": true, |
|
"language_model.model.layers.10.mlp.up_proj": true, |
|
"language_model.model.layers.10.input_layernorm": false, |
|
"language_model.model.layers.10.post_attention_layernorm": false, |
|
"language_model.model.layers.11.self_attn.q_proj": true, |
|
"language_model.model.layers.11.self_attn.k_proj": true, |
|
"language_model.model.layers.11.self_attn.v_proj": true, |
|
"language_model.model.layers.11.self_attn.o_proj": true, |
|
"language_model.model.layers.11.self_attn.rope": false, |
|
"language_model.model.layers.11.mlp.gate_proj": true, |
|
"language_model.model.layers.11.mlp.down_proj": true, |
|
"language_model.model.layers.11.mlp.up_proj": true, |
|
"language_model.model.layers.11.input_layernorm": false, |
|
"language_model.model.layers.11.post_attention_layernorm": false, |
|
"language_model.model.layers.12.self_attn.q_proj": true, |
|
"language_model.model.layers.12.self_attn.k_proj": true, |
|
"language_model.model.layers.12.self_attn.v_proj": true, |
|
"language_model.model.layers.12.self_attn.o_proj": true, |
|
"language_model.model.layers.12.self_attn.rope": false, |
|
"language_model.model.layers.12.mlp.gate_proj": true, |
|
"language_model.model.layers.12.mlp.down_proj": true, |
|
"language_model.model.layers.12.mlp.up_proj": true, |
|
"language_model.model.layers.12.input_layernorm": false, |
|
"language_model.model.layers.12.post_attention_layernorm": false, |
|
"language_model.model.layers.13.self_attn.q_proj": true, |
|
"language_model.model.layers.13.self_attn.k_proj": true, |
|
"language_model.model.layers.13.self_attn.v_proj": true, |
|
"language_model.model.layers.13.self_attn.o_proj": true, |
|
"language_model.model.layers.13.self_attn.rope": false, |
|
"language_model.model.layers.13.mlp.gate_proj": true, |
|
"language_model.model.layers.13.mlp.down_proj": true, |
|
"language_model.model.layers.13.mlp.up_proj": true, |
|
"language_model.model.layers.13.input_layernorm": false, |
|
"language_model.model.layers.13.post_attention_layernorm": false, |
|
"language_model.model.layers.14.self_attn.q_proj": true, |
|
"language_model.model.layers.14.self_attn.k_proj": true, |
|
"language_model.model.layers.14.self_attn.v_proj": true, |
|
"language_model.model.layers.14.self_attn.o_proj": true, |
|
"language_model.model.layers.14.self_attn.rope": false, |
|
"language_model.model.layers.14.mlp.gate_proj": true, |
|
"language_model.model.layers.14.mlp.down_proj": true, |
|
"language_model.model.layers.14.mlp.up_proj": true, |
|
"language_model.model.layers.14.input_layernorm": false, |
|
"language_model.model.layers.14.post_attention_layernorm": false, |
|
"language_model.model.layers.15.self_attn.q_proj": true, |
|
"language_model.model.layers.15.self_attn.k_proj": true, |
|
"language_model.model.layers.15.self_attn.v_proj": true, |
|
"language_model.model.layers.15.self_attn.o_proj": true, |
|
"language_model.model.layers.15.self_attn.rope": false, |
|
"language_model.model.layers.15.mlp.gate_proj": true, |
|
"language_model.model.layers.15.mlp.down_proj": true, |
|
"language_model.model.layers.15.mlp.up_proj": true, |
|
"language_model.model.layers.15.input_layernorm": false, |
|
"language_model.model.layers.15.post_attention_layernorm": false, |
|
"language_model.model.layers.16.self_attn.q_proj": true, |
|
"language_model.model.layers.16.self_attn.k_proj": true, |
|
"language_model.model.layers.16.self_attn.v_proj": true, |
|
"language_model.model.layers.16.self_attn.o_proj": true, |
|
"language_model.model.layers.16.self_attn.rope": false, |
|
"language_model.model.layers.16.mlp.gate_proj": true, |
|
"language_model.model.layers.16.mlp.down_proj": true, |
|
"language_model.model.layers.16.mlp.up_proj": true, |
|
"language_model.model.layers.16.input_layernorm": false, |
|
"language_model.model.layers.16.post_attention_layernorm": false, |
|
"language_model.model.layers.17.self_attn.q_proj": true, |
|
"language_model.model.layers.17.self_attn.k_proj": true, |
|
"language_model.model.layers.17.self_attn.v_proj": true, |
|
"language_model.model.layers.17.self_attn.o_proj": true, |
|
"language_model.model.layers.17.self_attn.rope": false, |
|
"language_model.model.layers.17.mlp.gate_proj": true, |
|
"language_model.model.layers.17.mlp.down_proj": true, |
|
"language_model.model.layers.17.mlp.up_proj": true, |
|
"language_model.model.layers.17.input_layernorm": false, |
|
"language_model.model.layers.17.post_attention_layernorm": false, |
|
"language_model.model.layers.18.self_attn.q_proj": true, |
|
"language_model.model.layers.18.self_attn.k_proj": true, |
|
"language_model.model.layers.18.self_attn.v_proj": true, |
|
"language_model.model.layers.18.self_attn.o_proj": true, |
|
"language_model.model.layers.18.self_attn.rope": false, |
|
"language_model.model.layers.18.mlp.gate_proj": true, |
|
"language_model.model.layers.18.mlp.down_proj": true, |
|
"language_model.model.layers.18.mlp.up_proj": true, |
|
"language_model.model.layers.18.input_layernorm": false, |
|
"language_model.model.layers.18.post_attention_layernorm": false, |
|
"language_model.model.layers.19.self_attn.q_proj": true, |
|
"language_model.model.layers.19.self_attn.k_proj": true, |
|
"language_model.model.layers.19.self_attn.v_proj": true, |
|
"language_model.model.layers.19.self_attn.o_proj": true, |
|
"language_model.model.layers.19.self_attn.rope": false, |
|
"language_model.model.layers.19.mlp.gate_proj": true, |
|
"language_model.model.layers.19.mlp.down_proj": true, |
|
"language_model.model.layers.19.mlp.up_proj": true, |
|
"language_model.model.layers.19.input_layernorm": false, |
|
"language_model.model.layers.19.post_attention_layernorm": false, |
|
"language_model.model.layers.20.self_attn.q_proj": true, |
|
"language_model.model.layers.20.self_attn.k_proj": true, |
|
"language_model.model.layers.20.self_attn.v_proj": true, |
|
"language_model.model.layers.20.self_attn.o_proj": true, |
|
"language_model.model.layers.20.self_attn.rope": false, |
|
"language_model.model.layers.20.mlp.gate_proj": true, |
|
"language_model.model.layers.20.mlp.down_proj": true, |
|
"language_model.model.layers.20.mlp.up_proj": true, |
|
"language_model.model.layers.20.input_layernorm": false, |
|
"language_model.model.layers.20.post_attention_layernorm": false, |
|
"language_model.model.layers.21.self_attn.q_proj": true, |
|
"language_model.model.layers.21.self_attn.k_proj": true, |
|
"language_model.model.layers.21.self_attn.v_proj": true, |
|
"language_model.model.layers.21.self_attn.o_proj": true, |
|
"language_model.model.layers.21.self_attn.rope": false, |
|
"language_model.model.layers.21.mlp.gate_proj": true, |
|
"language_model.model.layers.21.mlp.down_proj": true, |
|
"language_model.model.layers.21.mlp.up_proj": true, |
|
"language_model.model.layers.21.input_layernorm": false, |
|
"language_model.model.layers.21.post_attention_layernorm": false, |
|
"language_model.model.layers.22.self_attn.q_proj": true, |
|
"language_model.model.layers.22.self_attn.k_proj": true, |
|
"language_model.model.layers.22.self_attn.v_proj": true, |
|
"language_model.model.layers.22.self_attn.o_proj": true, |
|
"language_model.model.layers.22.self_attn.rope": false, |
|
"language_model.model.layers.22.mlp.gate_proj": true, |
|
"language_model.model.layers.22.mlp.down_proj": true, |
|
"language_model.model.layers.22.mlp.up_proj": true, |
|
"language_model.model.layers.22.input_layernorm": false, |
|
"language_model.model.layers.22.post_attention_layernorm": false, |
|
"language_model.model.layers.23.self_attn.q_proj": true, |
|
"language_model.model.layers.23.self_attn.k_proj": true, |
|
"language_model.model.layers.23.self_attn.v_proj": true, |
|
"language_model.model.layers.23.self_attn.o_proj": true, |
|
"language_model.model.layers.23.self_attn.rope": false, |
|
"language_model.model.layers.23.mlp.gate_proj": true, |
|
"language_model.model.layers.23.mlp.down_proj": true, |
|
"language_model.model.layers.23.mlp.up_proj": true, |
|
"language_model.model.layers.23.input_layernorm": false, |
|
"language_model.model.layers.23.post_attention_layernorm": false, |
|
"language_model.model.layers.24.self_attn.q_proj": true, |
|
"language_model.model.layers.24.self_attn.k_proj": true, |
|
"language_model.model.layers.24.self_attn.v_proj": true, |
|
"language_model.model.layers.24.self_attn.o_proj": true, |
|
"language_model.model.layers.24.self_attn.rope": false, |
|
"language_model.model.layers.24.mlp.gate_proj": true, |
|
"language_model.model.layers.24.mlp.down_proj": true, |
|
"language_model.model.layers.24.mlp.up_proj": true, |
|
"language_model.model.layers.24.input_layernorm": false, |
|
"language_model.model.layers.24.post_attention_layernorm": false, |
|
"language_model.model.layers.25.self_attn.q_proj": true, |
|
"language_model.model.layers.25.self_attn.k_proj": true, |
|
"language_model.model.layers.25.self_attn.v_proj": true, |
|
"language_model.model.layers.25.self_attn.o_proj": true, |
|
"language_model.model.layers.25.self_attn.rope": false, |
|
"language_model.model.layers.25.mlp.gate_proj": true, |
|
"language_model.model.layers.25.mlp.down_proj": true, |
|
"language_model.model.layers.25.mlp.up_proj": true, |
|
"language_model.model.layers.25.input_layernorm": false, |
|
"language_model.model.layers.25.post_attention_layernorm": false, |
|
"language_model.model.layers.26.self_attn.q_proj": true, |
|
"language_model.model.layers.26.self_attn.k_proj": true, |
|
"language_model.model.layers.26.self_attn.v_proj": true, |
|
"language_model.model.layers.26.self_attn.o_proj": true, |
|
"language_model.model.layers.26.self_attn.rope": false, |
|
"language_model.model.layers.26.mlp.gate_proj": true, |
|
"language_model.model.layers.26.mlp.down_proj": true, |
|
"language_model.model.layers.26.mlp.up_proj": true, |
|
"language_model.model.layers.26.input_layernorm": false, |
|
"language_model.model.layers.26.post_attention_layernorm": false, |
|
"language_model.model.layers.27.self_attn.q_proj": true, |
|
"language_model.model.layers.27.self_attn.k_proj": true, |
|
"language_model.model.layers.27.self_attn.v_proj": true, |
|
"language_model.model.layers.27.self_attn.o_proj": true, |
|
"language_model.model.layers.27.self_attn.rope": false, |
|
"language_model.model.layers.27.mlp.gate_proj": true, |
|
"language_model.model.layers.27.mlp.down_proj": true, |
|
"language_model.model.layers.27.mlp.up_proj": true, |
|
"language_model.model.layers.27.input_layernorm": false, |
|
"language_model.model.layers.27.post_attention_layernorm": false, |
|
"language_model.model.layers.28.self_attn.q_proj": true, |
|
"language_model.model.layers.28.self_attn.k_proj": true, |
|
"language_model.model.layers.28.self_attn.v_proj": true, |
|
"language_model.model.layers.28.self_attn.o_proj": true, |
|
"language_model.model.layers.28.self_attn.rope": false, |
|
"language_model.model.layers.28.mlp.gate_proj": true, |
|
"language_model.model.layers.28.mlp.down_proj": true, |
|
"language_model.model.layers.28.mlp.up_proj": true, |
|
"language_model.model.layers.28.input_layernorm": false, |
|
"language_model.model.layers.28.post_attention_layernorm": false, |
|
"language_model.model.layers.29.self_attn.q_proj": true, |
|
"language_model.model.layers.29.self_attn.k_proj": true, |
|
"language_model.model.layers.29.self_attn.v_proj": true, |
|
"language_model.model.layers.29.self_attn.o_proj": true, |
|
"language_model.model.layers.29.self_attn.rope": false, |
|
"language_model.model.layers.29.mlp.gate_proj": true, |
|
"language_model.model.layers.29.mlp.down_proj": true, |
|
"language_model.model.layers.29.mlp.up_proj": true, |
|
"language_model.model.layers.29.input_layernorm": false, |
|
"language_model.model.layers.29.post_attention_layernorm": false, |
|
"language_model.model.layers.30.self_attn.q_proj": true, |
|
"language_model.model.layers.30.self_attn.k_proj": true, |
|
"language_model.model.layers.30.self_attn.v_proj": true, |
|
"language_model.model.layers.30.self_attn.o_proj": true, |
|
"language_model.model.layers.30.self_attn.rope": false, |
|
"language_model.model.layers.30.mlp.gate_proj": true, |
|
"language_model.model.layers.30.mlp.down_proj": true, |
|
"language_model.model.layers.30.mlp.up_proj": true, |
|
"language_model.model.layers.30.input_layernorm": false, |
|
"language_model.model.layers.30.post_attention_layernorm": false, |
|
"language_model.model.layers.31.self_attn.q_proj": true, |
|
"language_model.model.layers.31.self_attn.k_proj": true, |
|
"language_model.model.layers.31.self_attn.v_proj": true, |
|
"language_model.model.layers.31.self_attn.o_proj": true, |
|
"language_model.model.layers.31.self_attn.rope": false, |
|
"language_model.model.layers.31.mlp.gate_proj": true, |
|
"language_model.model.layers.31.mlp.down_proj": true, |
|
"language_model.model.layers.31.mlp.up_proj": true, |
|
"language_model.model.layers.31.input_layernorm": false, |
|
"language_model.model.layers.31.post_attention_layernorm": false, |
|
"language_model.model.layers.32.self_attn.q_proj": true, |
|
"language_model.model.layers.32.self_attn.k_proj": true, |
|
"language_model.model.layers.32.self_attn.v_proj": true, |
|
"language_model.model.layers.32.self_attn.o_proj": true, |
|
"language_model.model.layers.32.self_attn.rope": false, |
|
"language_model.model.layers.32.mlp.gate_proj": true, |
|
"language_model.model.layers.32.mlp.down_proj": true, |
|
"language_model.model.layers.32.mlp.up_proj": true, |
|
"language_model.model.layers.32.input_layernorm": false, |
|
"language_model.model.layers.32.post_attention_layernorm": false, |
|
"language_model.model.layers.33.self_attn.q_proj": true, |
|
"language_model.model.layers.33.self_attn.k_proj": true, |
|
"language_model.model.layers.33.self_attn.v_proj": true, |
|
"language_model.model.layers.33.self_attn.o_proj": true, |
|
"language_model.model.layers.33.self_attn.rope": false, |
|
"language_model.model.layers.33.mlp.gate_proj": true, |
|
"language_model.model.layers.33.mlp.down_proj": true, |
|
"language_model.model.layers.33.mlp.up_proj": true, |
|
"language_model.model.layers.33.input_layernorm": false, |
|
"language_model.model.layers.33.post_attention_layernorm": false, |
|
"language_model.model.layers.34.self_attn.q_proj": true, |
|
"language_model.model.layers.34.self_attn.k_proj": true, |
|
"language_model.model.layers.34.self_attn.v_proj": true, |
|
"language_model.model.layers.34.self_attn.o_proj": true, |
|
"language_model.model.layers.34.self_attn.rope": false, |
|
"language_model.model.layers.34.mlp.gate_proj": true, |
|
"language_model.model.layers.34.mlp.down_proj": true, |
|
"language_model.model.layers.34.mlp.up_proj": true, |
|
"language_model.model.layers.34.input_layernorm": false, |
|
"language_model.model.layers.34.post_attention_layernorm": false, |
|
"language_model.model.layers.35.self_attn.q_proj": true, |
|
"language_model.model.layers.35.self_attn.k_proj": true, |
|
"language_model.model.layers.35.self_attn.v_proj": true, |
|
"language_model.model.layers.35.self_attn.o_proj": true, |
|
"language_model.model.layers.35.self_attn.rope": false, |
|
"language_model.model.layers.35.mlp.gate_proj": true, |
|
"language_model.model.layers.35.mlp.down_proj": true, |
|
"language_model.model.layers.35.mlp.up_proj": true, |
|
"language_model.model.layers.35.input_layernorm": false, |
|
"language_model.model.layers.35.post_attention_layernorm": false, |
|
"language_model.model.layers.36.self_attn.q_proj": true, |
|
"language_model.model.layers.36.self_attn.k_proj": true, |
|
"language_model.model.layers.36.self_attn.v_proj": true, |
|
"language_model.model.layers.36.self_attn.o_proj": true, |
|
"language_model.model.layers.36.self_attn.rope": false, |
|
"language_model.model.layers.36.mlp.gate_proj": true, |
|
"language_model.model.layers.36.mlp.down_proj": true, |
|
"language_model.model.layers.36.mlp.up_proj": true, |
|
"language_model.model.layers.36.input_layernorm": false, |
|
"language_model.model.layers.36.post_attention_layernorm": false, |
|
"language_model.model.layers.37.self_attn.q_proj": true, |
|
"language_model.model.layers.37.self_attn.k_proj": true, |
|
"language_model.model.layers.37.self_attn.v_proj": true, |
|
"language_model.model.layers.37.self_attn.o_proj": true, |
|
"language_model.model.layers.37.self_attn.rope": false, |
|
"language_model.model.layers.37.mlp.gate_proj": true, |
|
"language_model.model.layers.37.mlp.down_proj": true, |
|
"language_model.model.layers.37.mlp.up_proj": true, |
|
"language_model.model.layers.37.input_layernorm": false, |
|
"language_model.model.layers.37.post_attention_layernorm": false, |
|
"language_model.model.layers.38.self_attn.q_proj": true, |
|
"language_model.model.layers.38.self_attn.k_proj": true, |
|
"language_model.model.layers.38.self_attn.v_proj": true, |
|
"language_model.model.layers.38.self_attn.o_proj": true, |
|
"language_model.model.layers.38.self_attn.rope": false, |
|
"language_model.model.layers.38.mlp.gate_proj": true, |
|
"language_model.model.layers.38.mlp.down_proj": true, |
|
"language_model.model.layers.38.mlp.up_proj": true, |
|
"language_model.model.layers.38.input_layernorm": false, |
|
"language_model.model.layers.38.post_attention_layernorm": false, |
|
"language_model.model.layers.39.self_attn.q_proj": true, |
|
"language_model.model.layers.39.self_attn.k_proj": true, |
|
"language_model.model.layers.39.self_attn.v_proj": true, |
|
"language_model.model.layers.39.self_attn.o_proj": true, |
|
"language_model.model.layers.39.self_attn.rope": false, |
|
"language_model.model.layers.39.mlp.gate_proj": true, |
|
"language_model.model.layers.39.mlp.down_proj": true, |
|
"language_model.model.layers.39.mlp.up_proj": true, |
|
"language_model.model.layers.39.input_layernorm": false, |
|
"language_model.model.layers.39.post_attention_layernorm": false, |
|
"language_model.model.norm": false, |
|
"language_model.lm_head": true, |
|
"multi_modal_projector.norm": false, |
|
"multi_modal_projector.patch_merger.merging_layer": true, |
|
"multi_modal_projector.linear_1": true, |
|
"multi_modal_projector.gelu": false, |
|
"multi_modal_projector.linear_2": true |
|
}, |
|
"remove_invalid_values": false, |
|
"repetition_penalty": 1.0, |
|
"return_dict": true, |
|
"return_dict_in_generate": false, |
|
"sep_token_id": null, |
|
"spatial_merge_size": 2, |
|
"suppress_tokens": null, |
|
"task_specific_params": null, |
|
"temperature": 1.0, |
|
"text_config": { |
|
"vocab_size": 131072, |
|
"max_position_embeddings": 131072, |
|
"hidden_size": 5120, |
|
"intermediate_size": 32768, |
|
"num_hidden_layers": 40, |
|
"num_attention_heads": 32, |
|
"sliding_window": null, |
|
"head_dim": 128, |
|
"num_key_value_heads": 8, |
|
"hidden_act": "silu", |
|
"initializer_range": 0.02, |
|
"rms_norm_eps": 1e-05, |
|
"use_cache": true, |
|
"rope_theta": 1000000000.0, |
|
"attention_dropout": 0.0, |
|
"return_dict": true, |
|
"output_hidden_states": false, |
|
"torchscript": false, |
|
"torch_dtype": null, |
|
"use_bfloat16": false, |
|
"tf_legacy_loss": false, |
|
"pruned_heads": {}, |
|
"tie_word_embeddings": false, |
|
"chunk_size_feed_forward": 0, |
|
"is_encoder_decoder": false, |
|
"is_decoder": false, |
|
"cross_attention_hidden_size": null, |
|
"add_cross_attention": false, |
|
"tie_encoder_decoder": false, |
|
"max_length": 20, |
|
"min_length": 0, |
|
"do_sample": false, |
|
"early_stopping": false, |
|
"num_beams": 1, |
|
"num_beam_groups": 1, |
|
"diversity_penalty": 0.0, |
|
"temperature": 1.0, |
|
"top_k": 50, |
|
"top_p": 1.0, |
|
"typical_p": 1.0, |
|
"repetition_penalty": 1.0, |
|
"length_penalty": 1.0, |
|
"no_repeat_ngram_size": 0, |
|
"encoder_no_repeat_ngram_size": 0, |
|
"bad_words_ids": null, |
|
"num_return_sequences": 1, |
|
"output_scores": false, |
|
"return_dict_in_generate": false, |
|
"forced_bos_token_id": null, |
|
"forced_eos_token_id": null, |
|
"remove_invalid_values": false, |
|
"exponential_decay_length_penalty": null, |
|
"suppress_tokens": null, |
|
"begin_suppress_tokens": null, |
|
"architectures": null, |
|
"finetuning_task": null, |
|
"id2label": { |
|
"0": "LABEL_0", |
|
"1": "LABEL_1" |
|
}, |
|
"label2id": { |
|
"LABEL_0": 0, |
|
"LABEL_1": 1 |
|
}, |
|
"tokenizer_class": null, |
|
"prefix": null, |
|
"bos_token_id": 1, |
|
"pad_token_id": null, |
|
"eos_token_id": 2, |
|
"sep_token_id": null, |
|
"decoder_start_token_id": null, |
|
"task_specific_params": null, |
|
"problem_type": null, |
|
"_name_or_path": "", |
|
"model_type": "mistral", |
|
"output_attentions": false |
|
}, |
|
"tf_legacy_loss": false, |
|
"tie_encoder_decoder": false, |
|
"tie_word_embeddings": true, |
|
"tokenizer_class": null, |
|
"top_k": 50, |
|
"top_p": 1.0, |
|
"torchscript": false, |
|
"transformers_version": "4.53.2", |
|
"typical_p": 1.0, |
|
"use_bfloat16": false, |
|
"vision_config": { |
|
"return_dict": true, |
|
"output_hidden_states": false, |
|
"torchscript": false, |
|
"torch_dtype": null, |
|
"use_bfloat16": false, |
|
"tf_legacy_loss": false, |
|
"pruned_heads": {}, |
|
"tie_word_embeddings": true, |
|
"chunk_size_feed_forward": 0, |
|
"is_encoder_decoder": false, |
|
"is_decoder": false, |
|
"cross_attention_hidden_size": null, |
|
"add_cross_attention": false, |
|
"tie_encoder_decoder": false, |
|
"max_length": 20, |
|
"min_length": 0, |
|
"do_sample": false, |
|
"early_stopping": false, |
|
"num_beams": 1, |
|
"num_beam_groups": 1, |
|
"diversity_penalty": 0.0, |
|
"temperature": 1.0, |
|
"top_k": 50, |
|
"top_p": 1.0, |
|
"typical_p": 1.0, |
|
"repetition_penalty": 1.0, |
|
"length_penalty": 1.0, |
|
"no_repeat_ngram_size": 0, |
|
"encoder_no_repeat_ngram_size": 0, |
|
"bad_words_ids": null, |
|
"num_return_sequences": 1, |
|
"output_scores": false, |
|
"return_dict_in_generate": false, |
|
"forced_bos_token_id": null, |
|
"forced_eos_token_id": null, |
|
"remove_invalid_values": false, |
|
"exponential_decay_length_penalty": null, |
|
"suppress_tokens": null, |
|
"begin_suppress_tokens": null, |
|
"architectures": null, |
|
"finetuning_task": null, |
|
"id2label": { |
|
"0": "LABEL_0", |
|
"1": "LABEL_1" |
|
}, |
|
"label2id": { |
|
"LABEL_0": 0, |
|
"LABEL_1": 1 |
|
}, |
|
"tokenizer_class": null, |
|
"prefix": null, |
|
"bos_token_id": null, |
|
"pad_token_id": null, |
|
"eos_token_id": null, |
|
"sep_token_id": null, |
|
"decoder_start_token_id": null, |
|
"task_specific_params": null, |
|
"problem_type": null, |
|
"_name_or_path": "", |
|
"head_dim": 64, |
|
"model_type": "pixtral", |
|
"hidden_size": 1024, |
|
"intermediate_size": 4096, |
|
"num_hidden_layers": 24, |
|
"num_attention_heads": 16, |
|
"num_channels": 3, |
|
"patch_size": 14, |
|
"image_size": 1540, |
|
"attention_dropout": 0.0, |
|
"hidden_act": "silu", |
|
"rope_theta": 10000.0, |
|
"initializer_range": 0.02, |
|
"output_attentions": false |
|
}, |
|
"vision_feature_layer": -1 |
|
} |