{
"model_id": "videoalign",
"img_context_token_id": 151656,
"video_token_id": 151656,
"vision_start_token_id": 151652,
"output_dim": 1,
"reward_token": "special",
"loss_type": "btt",
"lora_apply_modules": ["text_decoder"],
"lora_mixed_training": true,
"lora_save_full_weight": true,
"image_encoder": {
"vision_encoder": {
"model_id": "videoalign_vit",
"num_layers": 32,
"hidden_size": 1280,
"ffn_hidden_size": 5120,
"llm_hidden_size": 1536,
"num_attention_heads": 16,
"hidden_dropout": 0.0,
"attention_dropout": 0.0,
"in_channels": 3,
"patch_size": 14,
"spatial_merge_size": 2,
"temporal_patch_size": 2,
"layernorm_epsilon": 1e-06,
"normalization": "LayerNorm",
"fp16": false,
"bf16": true,
"params_dtype": "bf16",
"activation_func": "quick_gelu",
"use_fused_rotary_pos_emb": false,
"post_layer_norm": false,
"pipeline_num_layers": [32],
"recompute_granularity": "full",
"recompute_method": "block",
"recompute_num_layers": 12
},
"vision_projector": {
"model_id": "lnmlp",
"num_layers": 1,
"num_attention_heads": 1,
"gated_linear_unit": false,
"bias_activation_fusion": false,
"add_bias_linear": true,
"input_size": 1280,
"hidden_size": 1536,
"ffn_hidden_size": 5120,
"activation_func": "gelu",
"bf16": true,
"params_dtype": "bf16",
"layernorm_epsilon": 1e-06,
"normalization": "LayerNorm"
}
},
"text_decoder": {
"model_id": "videoalign_lm",
"num_layers": 28,
"pipeline_num_layers": [28],
"hidden_size": 1536,
"ffn_hidden_size": 8960,
"num_attention_heads": 12,
"max_position_embeddings": 32768,
"vocab_size": 151660,
"rope_theta": 1000000.0,
"untie_embeddings_and_output_weights": false,
"disable_bias_linear": true,
"attention_dropout": 0.0,
"init_method_std": 0.01,
"hidden_dropout": 0.0,
"position_embedding_type": "mrope",
"normalization": "RMSNorm",
"activation_func": "silu",
"use_fused_rotary_pos_emb": true,
"use_remove_padding": true,
"sparse_mode": 3,
"attention_softmax_in_fp32": true,
"params_dtype": "bf16",
"bf16": true,
"parallel_output": true,
"group_query_attention": true,
"num_query_groups": 2,
"mrope_section": [16, 24, 24],
"rope_scaling": null,
"gated_linear_unit": true,
"layernorm_epsilon": 1e-06,
"add_bias_linear": false,
"add_qkv_bias": true,
"word_embeddings_training": true,
"word_embeddings_only_update_special": true
},
"text_encoder": null,
"video_encoder": null
}