Upload merged Qwen3-8B baseline and MKL3 S2 checkpoints
Browse files- .gitattributes +2 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/args.json +393 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/chat_template.jinja +1 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/chat_template.json +3 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/config.json +285 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/generation_config.json +12 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/model-00001-of-00004.safetensors +3 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/model-00002-of-00004.safetensors +3 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/model-00003-of-00004.safetensors +3 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/model-00004-of-00004.safetensors +3 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/model.safetensors.index.json +705 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/modeling_qwen3_5vit_qwen3.py +0 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/processing_qwen3_5vit_qwen3.py +302 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/processor_config.json +209 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/tokenizer.json +3 -0
- ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/tokenizer_config.json +19 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/args.json +393 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/chat_template.jinja +1 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/chat_template.json +3 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/config.json +289 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/generation_config.json +12 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/model-00001-of-00004.safetensors +3 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/model-00002-of-00004.safetensors +3 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/model-00003-of-00004.safetensors +3 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/model-00004-of-00004.safetensors +3 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/model.safetensors.index.json +720 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/modeling_qwen3_5vit_qwen3.py +0 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/processing_qwen3_5vit_qwen3.py +302 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/processor_config.json +209 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/tokenizer.json +3 -0
- ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/tokenizer_config.json +19 -0
.gitattributes
CHANGED
|
@@ -100,3 +100,5 @@ ckpts/S2/4b/radio_v3_large/radio_v3_large/v0-20261024-142322/checkpoint-1530/tok
|
|
| 100 |
ckpts/S2/4b/siglip2_so400m/siglip2_so400m/v0-20261024-102319/checkpoint-2182/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 101 |
ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 102 |
ckpts/S2/4b/siglip_hd/siglip_hd/v0-20261024-104810/checkpoint-1530/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
| 100 |
ckpts/S2/4b/siglip2_so400m/siglip2_so400m/v0-20261024-102319/checkpoint-2182/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 101 |
ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 102 |
ckpts/S2/4b/siglip_hd/siglip_hd/v0-20261024-104810/checkpoint-1530/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 103 |
+
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 104 |
+
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/args.json
ADDED
|
@@ -0,0 +1,393 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"output_dir": "/home/wangjunjie09/exps/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325",
|
| 3 |
+
"per_device_train_batch_size": 1,
|
| 4 |
+
"num_train_epochs": 1.0,
|
| 5 |
+
"max_steps": -1,
|
| 6 |
+
"learning_rate": 0.0001,
|
| 7 |
+
"lr_scheduler_type": "cosine",
|
| 8 |
+
"lr_scheduler_kwargs": null,
|
| 9 |
+
"warmup_steps": 0,
|
| 10 |
+
"optim": "adamw_torch_fused",
|
| 11 |
+
"optim_args": null,
|
| 12 |
+
"weight_decay": 0.1,
|
| 13 |
+
"adam_beta1": 0.9,
|
| 14 |
+
"adam_beta2": 0.95,
|
| 15 |
+
"adam_epsilon": 1e-08,
|
| 16 |
+
"optim_target_modules": null,
|
| 17 |
+
"gradient_accumulation_steps": 8,
|
| 18 |
+
"average_tokens_across_devices": true,
|
| 19 |
+
"max_grad_norm": 1.0,
|
| 20 |
+
"label_smoothing_factor": 0.0,
|
| 21 |
+
"bf16": true,
|
| 22 |
+
"fp16": false,
|
| 23 |
+
"bf16_full_eval": false,
|
| 24 |
+
"fp16_full_eval": false,
|
| 25 |
+
"tf32": null,
|
| 26 |
+
"gradient_checkpointing": true,
|
| 27 |
+
"gradient_checkpointing_kwargs": "{\"use_reentrant\": false}",
|
| 28 |
+
"torch_compile": false,
|
| 29 |
+
"torch_compile_backend": null,
|
| 30 |
+
"torch_compile_mode": null,
|
| 31 |
+
"use_liger_kernel": false,
|
| 32 |
+
"liger_kernel_config": null,
|
| 33 |
+
"use_cache": false,
|
| 34 |
+
"neftune_noise_alpha": null,
|
| 35 |
+
"torch_empty_cache_steps": null,
|
| 36 |
+
"auto_find_batch_size": false,
|
| 37 |
+
"logging_strategy": "steps",
|
| 38 |
+
"logging_steps": 5,
|
| 39 |
+
"logging_first_step": true,
|
| 40 |
+
"log_on_each_node": true,
|
| 41 |
+
"logging_nan_inf_filter": true,
|
| 42 |
+
"include_num_input_tokens_seen": false,
|
| 43 |
+
"log_level": "passive",
|
| 44 |
+
"log_level_replica": "warning",
|
| 45 |
+
"disable_tqdm": null,
|
| 46 |
+
"report_to": [
|
| 47 |
+
"none"
|
| 48 |
+
],
|
| 49 |
+
"run_name": "/home/wangjunjie09/exps/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325",
|
| 50 |
+
"project": "huggingface",
|
| 51 |
+
"trackio_space_id": "trackio",
|
| 52 |
+
"eval_strategy": "no",
|
| 53 |
+
"eval_steps": 100.0,
|
| 54 |
+
"eval_delay": 0,
|
| 55 |
+
"per_device_eval_batch_size": 1,
|
| 56 |
+
"prediction_loss_only": false,
|
| 57 |
+
"eval_on_start": false,
|
| 58 |
+
"eval_do_concat_batches": true,
|
| 59 |
+
"eval_use_gather_object": false,
|
| 60 |
+
"eval_accumulation_steps": null,
|
| 61 |
+
"include_for_metrics": [],
|
| 62 |
+
"batch_eval_metrics": false,
|
| 63 |
+
"save_only_model": false,
|
| 64 |
+
"save_strategy": "steps",
|
| 65 |
+
"save_steps": 100.0,
|
| 66 |
+
"save_on_each_node": false,
|
| 67 |
+
"save_total_limit": 1,
|
| 68 |
+
"enable_jit_checkpoint": false,
|
| 69 |
+
"push_to_hub": false,
|
| 70 |
+
"hub_token": null,
|
| 71 |
+
"hub_private_repo": null,
|
| 72 |
+
"hub_model_id": null,
|
| 73 |
+
"hub_strategy": "every_save",
|
| 74 |
+
"hub_always_push": false,
|
| 75 |
+
"hub_revision": null,
|
| 76 |
+
"load_best_model_at_end": false,
|
| 77 |
+
"metric_for_best_model": "loss",
|
| 78 |
+
"greater_is_better": false,
|
| 79 |
+
"ignore_data_skip": false,
|
| 80 |
+
"restore_callback_states_from_checkpoint": false,
|
| 81 |
+
"full_determinism": false,
|
| 82 |
+
"seed": 42,
|
| 83 |
+
"data_seed": 42,
|
| 84 |
+
"use_cpu": false,
|
| 85 |
+
"accelerator_config": "{\"dispatch_batches\": false}",
|
| 86 |
+
"parallelism_config": null,
|
| 87 |
+
"dataloader_drop_last": false,
|
| 88 |
+
"dataloader_num_workers": 8,
|
| 89 |
+
"dataloader_pin_memory": true,
|
| 90 |
+
"dataloader_persistent_workers": true,
|
| 91 |
+
"dataloader_prefetch_factor": 4,
|
| 92 |
+
"remove_unused_columns": true,
|
| 93 |
+
"label_names": null,
|
| 94 |
+
"train_sampling_strategy": "random",
|
| 95 |
+
"length_column_name": "length",
|
| 96 |
+
"ddp_find_unused_parameters": null,
|
| 97 |
+
"ddp_bucket_cap_mb": null,
|
| 98 |
+
"ddp_broadcast_buffers": null,
|
| 99 |
+
"ddp_backend": null,
|
| 100 |
+
"ddp_timeout": 7200,
|
| 101 |
+
"fsdp": [],
|
| 102 |
+
"fsdp_config": null,
|
| 103 |
+
"deepspeed": {
|
| 104 |
+
"fp16": {
|
| 105 |
+
"enabled": "auto",
|
| 106 |
+
"loss_scale": 0,
|
| 107 |
+
"loss_scale_window": 1000,
|
| 108 |
+
"initial_scale_power": 16,
|
| 109 |
+
"hysteresis": 2,
|
| 110 |
+
"min_loss_scale": 1
|
| 111 |
+
},
|
| 112 |
+
"bf16": {
|
| 113 |
+
"enabled": "auto"
|
| 114 |
+
},
|
| 115 |
+
"zero_optimization": {
|
| 116 |
+
"stage": 2,
|
| 117 |
+
"offload_optimizer": {
|
| 118 |
+
"device": "none",
|
| 119 |
+
"pin_memory": true
|
| 120 |
+
},
|
| 121 |
+
"allgather_partitions": true,
|
| 122 |
+
"allgather_bucket_size": 200000000.0,
|
| 123 |
+
"overlap_comm": false,
|
| 124 |
+
"reduce_scatter": true,
|
| 125 |
+
"reduce_bucket_size": 200000000.0,
|
| 126 |
+
"contiguous_gradients": true
|
| 127 |
+
},
|
| 128 |
+
"gradient_accumulation_steps": "auto",
|
| 129 |
+
"gradient_clipping": "auto",
|
| 130 |
+
"steps_per_print": 2000,
|
| 131 |
+
"train_batch_size": "auto",
|
| 132 |
+
"train_micro_batch_size_per_gpu": "auto",
|
| 133 |
+
"wall_clock_breakdown": false
|
| 134 |
+
},
|
| 135 |
+
"debug": null,
|
| 136 |
+
"skip_memory_metrics": true,
|
| 137 |
+
"do_train": false,
|
| 138 |
+
"do_eval": false,
|
| 139 |
+
"do_predict": false,
|
| 140 |
+
"resume_from_checkpoint": null,
|
| 141 |
+
"warmup_ratio": 0.05,
|
| 142 |
+
"logging_dir": "/home/wangjunjie09/exps/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/runs",
|
| 143 |
+
"local_rank": 0,
|
| 144 |
+
"sortish_sampler": false,
|
| 145 |
+
"predict_with_generate": false,
|
| 146 |
+
"generation_max_length": null,
|
| 147 |
+
"generation_num_beams": null,
|
| 148 |
+
"generation_config": null,
|
| 149 |
+
"tuner_backend": "peft",
|
| 150 |
+
"llm_checkpointing": 14,
|
| 151 |
+
"vit_gradient_checkpointing": true,
|
| 152 |
+
"router_aux_loss_coef": 0.0,
|
| 153 |
+
"enable_dft_loss": false,
|
| 154 |
+
"enable_channel_loss": false,
|
| 155 |
+
"safe_serialization": true,
|
| 156 |
+
"max_shard_size": "5GB",
|
| 157 |
+
"check_model": true,
|
| 158 |
+
"acc_strategy": "token",
|
| 159 |
+
"train_dataloader_shuffle": true,
|
| 160 |
+
"group_by_length": false,
|
| 161 |
+
"group_by_packed_video_cost": true,
|
| 162 |
+
"group_by_packed_compute_cost": false,
|
| 163 |
+
"group_by_packed_modality": false,
|
| 164 |
+
"max_epochs": null,
|
| 165 |
+
"aligner_lr": 1e-05,
|
| 166 |
+
"vit_lr": 1e-06,
|
| 167 |
+
"use_logits_to_keep": true,
|
| 168 |
+
"ds3_gather_for_generation": true,
|
| 169 |
+
"resume_only_model": false,
|
| 170 |
+
"optimizer": null,
|
| 171 |
+
"loss_type": null,
|
| 172 |
+
"eval_metric": null,
|
| 173 |
+
"callbacks": [],
|
| 174 |
+
"early_stop_interval": null,
|
| 175 |
+
"eval_use_evalscope": false,
|
| 176 |
+
"eval_dataset": [],
|
| 177 |
+
"eval_dataset_args": null,
|
| 178 |
+
"eval_limit": null,
|
| 179 |
+
"eval_generation_config": null,
|
| 180 |
+
"extra_eval_args": null,
|
| 181 |
+
"tuner_type": "lora",
|
| 182 |
+
"use_galore": false,
|
| 183 |
+
"galore_target_modules": null,
|
| 184 |
+
"galore_rank": 128,
|
| 185 |
+
"galore_update_proj_gap": 50,
|
| 186 |
+
"galore_scale": 1.0,
|
| 187 |
+
"galore_proj_type": "std",
|
| 188 |
+
"galore_optim_per_parameter": false,
|
| 189 |
+
"galore_with_embedding": false,
|
| 190 |
+
"galore_quantization": false,
|
| 191 |
+
"galore_proj_quant": false,
|
| 192 |
+
"galore_proj_bits": 4,
|
| 193 |
+
"galore_proj_group_size": 256,
|
| 194 |
+
"galore_cos_threshold": 0.4,
|
| 195 |
+
"galore_gamma_proj": 2,
|
| 196 |
+
"galore_queue_size": 5,
|
| 197 |
+
"lisa_activated_layers": 0,
|
| 198 |
+
"lisa_step_interval": 20,
|
| 199 |
+
"use_flash_ckpt": false,
|
| 200 |
+
"use_ray": false,
|
| 201 |
+
"ray_exp_name": null,
|
| 202 |
+
"device_groups": null,
|
| 203 |
+
"model": "/home/wangjunjie09/exps/S1/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261020-120802/checkpoint-2181",
|
| 204 |
+
"model_type": "llava_qwen3_5vit_qwen3",
|
| 205 |
+
"model_revision": null,
|
| 206 |
+
"task_type": "causal_lm",
|
| 207 |
+
"torch_dtype": "bfloat16",
|
| 208 |
+
"attn_impl": "flash_attn",
|
| 209 |
+
"experts_impl": null,
|
| 210 |
+
"new_special_tokens": [],
|
| 211 |
+
"num_labels": null,
|
| 212 |
+
"problem_type": null,
|
| 213 |
+
"rope_scaling": null,
|
| 214 |
+
"device_map": null,
|
| 215 |
+
"max_memory": {},
|
| 216 |
+
"max_model_len": null,
|
| 217 |
+
"local_repo_path": null,
|
| 218 |
+
"init_strategy": null,
|
| 219 |
+
"template": "llava_qwen3_5vit_qwen3_nonthinking",
|
| 220 |
+
"system": null,
|
| 221 |
+
"max_length": 16384,
|
| 222 |
+
"truncation_strategy": "delete",
|
| 223 |
+
"max_pixels": null,
|
| 224 |
+
"agent_template": null,
|
| 225 |
+
"norm_bbox": null,
|
| 226 |
+
"use_chat_template": true,
|
| 227 |
+
"padding_side": "right",
|
| 228 |
+
"padding_free": true,
|
| 229 |
+
"loss_scale": "default+ignore_empty_think",
|
| 230 |
+
"sequence_parallel_size": 1,
|
| 231 |
+
"template_backend": "swift",
|
| 232 |
+
"response_prefix": null,
|
| 233 |
+
"enable_thinking": false,
|
| 234 |
+
"add_non_thinking_prefix": true,
|
| 235 |
+
"dataset": [],
|
| 236 |
+
"val_dataset": [],
|
| 237 |
+
"cached_dataset": [
|
| 238 |
+
"/share/m2v_intern/wangjunjie09/datasets/prepared/s2/qwen3_8b/image/train",
|
| 239 |
+
"/share/m2v_intern/wangjunjie09/datasets/prepared/s2/qwen3_8b/video/train"
|
| 240 |
+
],
|
| 241 |
+
"cached_val_dataset": [],
|
| 242 |
+
"split_dataset_ratio": 0.0,
|
| 243 |
+
"dataset_num_proc": 16,
|
| 244 |
+
"load_from_cache_file": false,
|
| 245 |
+
"dataset_shuffle": true,
|
| 246 |
+
"val_dataset_shuffle": false,
|
| 247 |
+
"streaming": false,
|
| 248 |
+
"interleave_prob": null,
|
| 249 |
+
"stopping_strategy": "first_exhausted",
|
| 250 |
+
"shuffle_buffer_size": 1000,
|
| 251 |
+
"download_mode": "reuse_dataset_if_exists",
|
| 252 |
+
"columns": {},
|
| 253 |
+
"strict": false,
|
| 254 |
+
"model_name": null,
|
| 255 |
+
"model_author": null,
|
| 256 |
+
"custom_dataset_info": [],
|
| 257 |
+
"quant_method": null,
|
| 258 |
+
"quant_bits": null,
|
| 259 |
+
"hqq_axis": null,
|
| 260 |
+
"bnb_4bit_compute_dtype": "bfloat16",
|
| 261 |
+
"bnb_4bit_quant_type": "nf4",
|
| 262 |
+
"bnb_4bit_use_double_quant": true,
|
| 263 |
+
"bnb_4bit_quant_storage": null,
|
| 264 |
+
"max_new_tokens": 64,
|
| 265 |
+
"temperature": 0.0,
|
| 266 |
+
"top_k": null,
|
| 267 |
+
"top_p": null,
|
| 268 |
+
"repetition_penalty": null,
|
| 269 |
+
"num_beams": 1,
|
| 270 |
+
"stream": false,
|
| 271 |
+
"stop_words": [],
|
| 272 |
+
"logprobs": false,
|
| 273 |
+
"top_logprobs": null,
|
| 274 |
+
"structured_outputs_regex": null,
|
| 275 |
+
"train_type": null,
|
| 276 |
+
"adapters": [],
|
| 277 |
+
"external_plugins": [
|
| 278 |
+
"video_mllm/training/model_plugin_qwen3_5vit.py",
|
| 279 |
+
"video_mllm/data/dataset_plugin.py",
|
| 280 |
+
"/home/wangjunjie09/VisionEncoder/video_mllm/training/log_spacing_plugin.py"
|
| 281 |
+
],
|
| 282 |
+
"custom_register_path": [
|
| 283 |
+
"/home/wangjunjie09/VisionEncoder/video_mllm/training/log_spacing_plugin.py"
|
| 284 |
+
],
|
| 285 |
+
"model_kwargs": {
|
| 286 |
+
"vision_source_id": "qwen3_5_2b",
|
| 287 |
+
"qwen3_5_vit_video_mode": "temporal_merge",
|
| 288 |
+
"num_frames": 64,
|
| 289 |
+
"eval_frames": 64,
|
| 290 |
+
"s2_use_video_frame_cache": true,
|
| 291 |
+
"s2_video_frame_cache": "/share/m2v_intern/wangjunjie09/datasets/video_cache/s2/frames/jpeg/384/64"
|
| 292 |
+
},
|
| 293 |
+
"video_cache": {},
|
| 294 |
+
"image_teacher_models": {},
|
| 295 |
+
"load_args": false,
|
| 296 |
+
"load_data_args": false,
|
| 297 |
+
"packing": true,
|
| 298 |
+
"packing_length": 16384,
|
| 299 |
+
"packing_num_proc": 1,
|
| 300 |
+
"lazy_tokenize": false,
|
| 301 |
+
"use_hf": true,
|
| 302 |
+
"ignore_args_error": false,
|
| 303 |
+
"use_swift_lora": false,
|
| 304 |
+
"freeze_parameters": [],
|
| 305 |
+
"freeze_parameters_regex": null,
|
| 306 |
+
"freeze_parameters_ratio": 0.0,
|
| 307 |
+
"trainable_parameters": [],
|
| 308 |
+
"trainable_parameters_regex": null,
|
| 309 |
+
"freeze_llm": false,
|
| 310 |
+
"freeze_vit": false,
|
| 311 |
+
"freeze_aligner": false,
|
| 312 |
+
"target_modules": "model\\.language_model\\.layers\\.[0-9]+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))",
|
| 313 |
+
"target_regex": "model\\.language_model\\.layers\\.[0-9]+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))",
|
| 314 |
+
"target_parameters": null,
|
| 315 |
+
"modules_to_save": [
|
| 316 |
+
"vision_tower",
|
| 317 |
+
"multi_modal_projector"
|
| 318 |
+
],
|
| 319 |
+
"lora_rank": 128,
|
| 320 |
+
"lora_alpha": 256,
|
| 321 |
+
"lora_dropout": 0.0,
|
| 322 |
+
"lora_bias": "none",
|
| 323 |
+
"lora_dtype": null,
|
| 324 |
+
"lorap_lr_ratio": null,
|
| 325 |
+
"use_rslora": false,
|
| 326 |
+
"use_dora": false,
|
| 327 |
+
"lora_ga_batch_size": 2,
|
| 328 |
+
"lora_ga_iters": 2,
|
| 329 |
+
"lora_ga_max_length": 1024,
|
| 330 |
+
"lora_ga_direction": "ArB2r",
|
| 331 |
+
"lora_ga_scale": "stable",
|
| 332 |
+
"lora_ga_stable_gamma": 16,
|
| 333 |
+
"init_weights": true,
|
| 334 |
+
"fourier_n_frequency": 2000,
|
| 335 |
+
"fourier_scaling": 300.0,
|
| 336 |
+
"boft_block_size": 4,
|
| 337 |
+
"boft_block_num": 0,
|
| 338 |
+
"boft_n_butterfly_factor": 1,
|
| 339 |
+
"boft_dropout": 0.0,
|
| 340 |
+
"vera_rank": 256,
|
| 341 |
+
"vera_projection_prng_key": 0,
|
| 342 |
+
"vera_dropout": 0.0,
|
| 343 |
+
"vera_d_initial": 0.1,
|
| 344 |
+
"adapter_act": "gelu",
|
| 345 |
+
"adapter_length": 128,
|
| 346 |
+
"adalora_target_r": 8,
|
| 347 |
+
"adalora_init_r": 12,
|
| 348 |
+
"adalora_tinit": 0,
|
| 349 |
+
"adalora_tfinal": 0,
|
| 350 |
+
"adalora_deltaT": 1,
|
| 351 |
+
"adalora_beta1": 0.85,
|
| 352 |
+
"adalora_beta2": 0.85,
|
| 353 |
+
"adalora_orth_reg_weight": 0.5,
|
| 354 |
+
"llamapro_num_new_blocks": 4,
|
| 355 |
+
"llamapro_num_groups": null,
|
| 356 |
+
"reft_layer_key": null,
|
| 357 |
+
"reft_layers": null,
|
| 358 |
+
"reft_rank": 4,
|
| 359 |
+
"reft_intervention_type": "LoreftIntervention",
|
| 360 |
+
"reft_args": null,
|
| 361 |
+
"swanlab_token": null,
|
| 362 |
+
"swanlab_project": "ms-swift",
|
| 363 |
+
"swanlab_workspace": null,
|
| 364 |
+
"swanlab_exp_name": null,
|
| 365 |
+
"swanlab_notification_method": null,
|
| 366 |
+
"swanlab_webhook_url": null,
|
| 367 |
+
"swanlab_secret": null,
|
| 368 |
+
"swanlab_sender_email": null,
|
| 369 |
+
"swanlab_receiver_email": null,
|
| 370 |
+
"swanlab_smtp_server": null,
|
| 371 |
+
"swanlab_smtp_port": null,
|
| 372 |
+
"swanlab_email_language": "zh",
|
| 373 |
+
"swanlab_mode": "cloud",
|
| 374 |
+
"add_version": true,
|
| 375 |
+
"create_checkpoint_symlink": false,
|
| 376 |
+
"v12_resolution": 384,
|
| 377 |
+
"zero_hpz_partition_size": null,
|
| 378 |
+
"deepspeed_autotp_size": null,
|
| 379 |
+
"swift_version": "4.1.0.dev0",
|
| 380 |
+
"ckpt_dir": "/home/wangjunjie09/exps/S1/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261020-120802/checkpoint-2181",
|
| 381 |
+
"rank": 0,
|
| 382 |
+
"global_world_size": 8,
|
| 383 |
+
"local_world_size": 8,
|
| 384 |
+
"model_suffix": "checkpoint-2181",
|
| 385 |
+
"model_info": "ModelInfo(model_type='llava_qwen3_5vit_qwen3', model_dir='/home/wangjunjie09/exps/S1/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261020-120802/checkpoint-2181', torch_dtype=torch.bfloat16, max_model_len=40960, quant_method=None, quant_bits=None, rope_scaling=None, is_moe_model=False, is_multimodal=True, config=None, task_type='causal_lm', num_labels=None)",
|
| 386 |
+
"model_meta": "ModelMeta(model_type='llava_qwen3_5vit_qwen3', model_groups=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen3-0.6B', hf_model_id='Qwen/Qwen3-0.6B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-1.7B', hf_model_id='Qwen/Qwen3-1.7B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-4B', hf_model_id='Qwen/Qwen3-4B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-8B', hf_model_id='Qwen/Qwen3-8B', model_path=None, ms_revision=None, hf_revision=None)], template=None, ignore_patterns=None, requires=None, tags=[])], loader=<class 'model_plugin_qwen3_5vit.Qwen3_5ViTLoader'>, template='llava_qwen3_5vit_qwen3', model_arch=MultiModelKeys(arch_name='llava_hf', embedding=None, module_list=None, lm_head=None, q_proj=None, k_proj=None, v_proj=None, o_proj=None, attention=None, mlp=None, down_proj=None, qkv_proj=None, qk_proj=None, qa_proj=None, qb_proj=None, kv_proj=None, kva_proj=None, kvb_proj=None, language_model=['model.language_model', 'lm_head'], aligner=['model.multi_modal_projector'], vision_tower=['model.vision_tower'], generator=[]), architectures=['LlavaQwen3_5ViTForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=[], tags=['vision', 'video'])",
|
| 387 |
+
"model_dir": "/home/wangjunjie09/exps/S1/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261020-120802/checkpoint-2181",
|
| 388 |
+
"template_meta": "Qwen3MixedTemplateMeta(template_type='llava_qwen3_5vit_qwen3_nonthinking', prefix=[], prompt=['<|im_start|>user\\n{{QUERY}}<|im_end|>\\n<|im_start|>assistant\\n'], chat_sep=['<|im_end|>\\n'], suffix=['<|im_end|>\\n'], template_cls=<class 'video_mllm.training.template_qwen3_5vit.Qwen3_5ViTLlavaTemplate'>, system_prefix=['<|im_start|>system\\n{{SYSTEM}}<|im_end|>\\n'], default_system=None, auto_add_bos=False, stop_words=['<|endoftext|>'], agent_template='hermes', is_thinking=False, thinking_prefix='', non_thinking_prefix='<think>\\n\\n</think>\\n\\n', history_thinking_prefix='')",
|
| 389 |
+
"_val_dataset_exists": false,
|
| 390 |
+
"hub": "<class 'swift.hub.hub.HFHub'>",
|
| 391 |
+
"evaluation_strategy": "steps",
|
| 392 |
+
"training_args": "Seq2SeqTrainingArguments(output_dir='/home/wangjunjie09/exps/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325', per_device_train_batch_size=1, num_train_epochs=1.0, max_steps=-1, learning_rate=0.0001, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_steps=0.05, optim=<OptimizerNames.ADAMW_TORCH_FUSED: 'adamw_torch_fused'>, optim_args=None, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, optim_target_modules=None, gradient_accumulation_steps=8, average_tokens_across_devices=None, max_grad_norm=1.0, label_smoothing_factor=0.0, bf16=True, fp16=False, bf16_full_eval=False, fp16_full_eval=False, tf32=None, gradient_checkpointing=True, gradient_checkpointing_kwargs={'use_reentrant': False}, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, use_liger_kernel=False, liger_kernel_config=None, use_cache=False, neftune_noise_alpha=None, torch_empty_cache_steps=None, auto_find_batch_size=False, logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_steps=5, logging_first_step=True, log_on_each_node=True, logging_nan_inf_filter=True, include_num_input_tokens_seen=None, log_level='passive', log_level_replica='warning', disable_tqdm=False, report_to=[], run_name='/home/wangjunjie09/exps/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325', project='huggingface', trackio_space_id='trackio', eval_strategy=<IntervalStrategy.NO: 'no'>, eval_steps=100.0, eval_delay=0, per_device_eval_batch_size=1, prediction_loss_only=False, eval_on_start=False, eval_do_concat_batches=True, eval_use_gather_object=False, eval_accumulation_steps=None, include_for_metrics=[], batch_eval_metrics=False, save_only_model=False, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=100, save_on_each_node=False, save_total_limit=1, enable_jit_checkpoint=False, push_to_hub=False, hub_token=None, hub_private_repo=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_always_push=False, hub_revision=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, restore_callback_states_from_checkpoint=False, full_determinism=False, seed=42, data_seed=42, use_cpu=False, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), parallelism_config=None, dataloader_drop_last=False, dataloader_num_workers=8, dataloader_pin_memory=True, dataloader_persistent_workers=True, dataloader_prefetch_factor=4, remove_unused_columns=False, label_names=None, train_sampling_strategy='random', length_column_name='length', ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, ddp_backend=None, ddp_timeout=7200, fsdp=[], fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, deepspeed={'fp16': {'enabled': 'auto', 'loss_scale': 0, 'loss_scale_window': 1000, 'initial_scale_power': 16, 'hysteresis': 2, 'min_loss_scale': 1}, 'bf16': {'enabled': 'auto'}, 'zero_optimization': {'stage': 2, 'offload_optimizer': {'device': 'none', 'pin_memory': True}, 'allgather_partitions': True, 'allgather_bucket_size': 200000000.0, 'overlap_comm': False, 'reduce_scatter': True, 'reduce_bucket_size': 200000000.0, 'contiguous_gradients': True}, 'gradient_accumulation_steps': 'auto', 'gradient_clipping': 'auto', 'steps_per_print': 2000, 'train_batch_size': 'auto', 'train_micro_batch_size_per_gpu': 'auto', 'wall_clock_breakdown': False}, debug=[], skip_memory_metrics=True, do_train=False, do_eval=False, do_predict=False, resume_from_checkpoint=None, warmup_ratio=0.05, logging_dir='/home/wangjunjie09/exps/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/runs', local_rank=0, sortish_sampler=False, predict_with_generate=False, generation_max_length=None, generation_num_beams=None, generation_config=None, tuner_backend='peft', llm_checkpointing=14, vit_gradient_checkpointing=True, router_aux_loss_coef=0.0, enable_dft_loss=False, enable_channel_loss=False, safe_serialization=True, max_shard_size='5GB', check_model=True, acc_strategy='token', train_dataloader_shuffle=True, group_by_length=False, group_by_packed_video_cost=True, group_by_packed_compute_cost=False, group_by_packed_modality=False, max_epochs=None, aligner_lr=1e-05, vit_lr=1e-06, use_logits_to_keep=True, ds3_gather_for_generation=True, resume_only_model=False, optimizer='multimodal', loss_type=None, eval_metric=None, callbacks=[], early_stop_interval=None, eval_use_evalscope=False, eval_dataset=[], eval_dataset_args=None, eval_limit=None, eval_generation_config=None, extra_eval_args=None, tuner_type='lora', use_galore=False, galore_target_modules=None, galore_rank=128, galore_update_proj_gap=50, galore_scale=1.0, galore_proj_type='std', galore_optim_per_parameter=False, galore_with_embedding=False, galore_quantization=False, galore_proj_quant=False, galore_proj_bits=4, galore_proj_group_size=256, galore_cos_threshold=0.4, galore_gamma_proj=2, galore_queue_size=5, lisa_activated_layers=0, lisa_step_interval=20, use_flash_ckpt=False)"
|
| 393 |
+
}
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/chat_template.jinja
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{% for message in messages %}{{ '<|im_start|>' + message['role'] + '\n' }}{% if message['role'] == 'assistant' %}{{ '<think>\n\n</think>\n\n' }}{% endif %}{% if message['content'] is string %}{{ message['content'] }}{% else %}{% for content in message['content'] | selectattr('type', 'equalto', 'image') %}{{ '<image>\n' }}{% endfor %}{% for content in message['content'] | selectattr('type', 'equalto', 'video') %}{{ '<video>\n' }}{% endfor %}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{{ content['text'] }}{% endfor %}{% endif %}{{ '<|im_end|>\n' }}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n<think>\n\n</think>\n\n' }}{% endif %}
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/chat_template.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"chat_template": "{% for message in messages %}{{ '<|im_start|>' + message['role'] + '\\n' }}{% if message['role'] == 'assistant' %}{{ '<think>\\n\\n</think>\\n\\n' }}{% endif %}{% if message['content'] is string %}{{ message['content'] }}{% else %}{% for content in message['content'] | selectattr('type', 'equalto', 'image') %}{{ '<image>\\n' }}{% endfor %}{% for content in message['content'] | selectattr('type', 'equalto', 'video') %}{{ '<video>\\n' }}{% endfor %}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{{ content['text'] }}{% endfor %}{% endif %}{{ '<|im_end|>\\n' }}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\\n<think>\\n\\n</think>\\n\\n' }}{% endif %}"
|
| 3 |
+
}
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/config.json
ADDED
|
@@ -0,0 +1,285 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"LlavaQwen3_5ViTForConditionalGeneration"
|
| 4 |
+
],
|
| 5 |
+
"bos_token_id": null,
|
| 6 |
+
"dtype": "bfloat16",
|
| 7 |
+
"eos_token_id": 151645,
|
| 8 |
+
"eval_num_frames": 64,
|
| 9 |
+
"hidden_size": 4096,
|
| 10 |
+
"image_grid_pinpoints": [
|
| 11 |
+
[
|
| 12 |
+
384,
|
| 13 |
+
384
|
| 14 |
+
],
|
| 15 |
+
[
|
| 16 |
+
384,
|
| 17 |
+
768
|
| 18 |
+
],
|
| 19 |
+
[
|
| 20 |
+
384,
|
| 21 |
+
1152
|
| 22 |
+
],
|
| 23 |
+
[
|
| 24 |
+
384,
|
| 25 |
+
1536
|
| 26 |
+
],
|
| 27 |
+
[
|
| 28 |
+
384,
|
| 29 |
+
1920
|
| 30 |
+
],
|
| 31 |
+
[
|
| 32 |
+
384,
|
| 33 |
+
2304
|
| 34 |
+
],
|
| 35 |
+
[
|
| 36 |
+
768,
|
| 37 |
+
384
|
| 38 |
+
],
|
| 39 |
+
[
|
| 40 |
+
768,
|
| 41 |
+
768
|
| 42 |
+
],
|
| 43 |
+
[
|
| 44 |
+
768,
|
| 45 |
+
1152
|
| 46 |
+
],
|
| 47 |
+
[
|
| 48 |
+
768,
|
| 49 |
+
1536
|
| 50 |
+
],
|
| 51 |
+
[
|
| 52 |
+
768,
|
| 53 |
+
1920
|
| 54 |
+
],
|
| 55 |
+
[
|
| 56 |
+
768,
|
| 57 |
+
2304
|
| 58 |
+
],
|
| 59 |
+
[
|
| 60 |
+
1152,
|
| 61 |
+
384
|
| 62 |
+
],
|
| 63 |
+
[
|
| 64 |
+
1152,
|
| 65 |
+
768
|
| 66 |
+
],
|
| 67 |
+
[
|
| 68 |
+
1152,
|
| 69 |
+
1152
|
| 70 |
+
],
|
| 71 |
+
[
|
| 72 |
+
1152,
|
| 73 |
+
1536
|
| 74 |
+
],
|
| 75 |
+
[
|
| 76 |
+
1152,
|
| 77 |
+
1920
|
| 78 |
+
],
|
| 79 |
+
[
|
| 80 |
+
1152,
|
| 81 |
+
2304
|
| 82 |
+
],
|
| 83 |
+
[
|
| 84 |
+
1536,
|
| 85 |
+
384
|
| 86 |
+
],
|
| 87 |
+
[
|
| 88 |
+
1536,
|
| 89 |
+
768
|
| 90 |
+
],
|
| 91 |
+
[
|
| 92 |
+
1536,
|
| 93 |
+
1152
|
| 94 |
+
],
|
| 95 |
+
[
|
| 96 |
+
1536,
|
| 97 |
+
1536
|
| 98 |
+
],
|
| 99 |
+
[
|
| 100 |
+
1536,
|
| 101 |
+
1920
|
| 102 |
+
],
|
| 103 |
+
[
|
| 104 |
+
1536,
|
| 105 |
+
2304
|
| 106 |
+
],
|
| 107 |
+
[
|
| 108 |
+
1920,
|
| 109 |
+
384
|
| 110 |
+
],
|
| 111 |
+
[
|
| 112 |
+
1920,
|
| 113 |
+
768
|
| 114 |
+
],
|
| 115 |
+
[
|
| 116 |
+
1920,
|
| 117 |
+
1152
|
| 118 |
+
],
|
| 119 |
+
[
|
| 120 |
+
1920,
|
| 121 |
+
1536
|
| 122 |
+
],
|
| 123 |
+
[
|
| 124 |
+
1920,
|
| 125 |
+
1920
|
| 126 |
+
],
|
| 127 |
+
[
|
| 128 |
+
1920,
|
| 129 |
+
2304
|
| 130 |
+
],
|
| 131 |
+
[
|
| 132 |
+
2304,
|
| 133 |
+
384
|
| 134 |
+
],
|
| 135 |
+
[
|
| 136 |
+
2304,
|
| 137 |
+
768
|
| 138 |
+
],
|
| 139 |
+
[
|
| 140 |
+
2304,
|
| 141 |
+
1152
|
| 142 |
+
],
|
| 143 |
+
[
|
| 144 |
+
2304,
|
| 145 |
+
1536
|
| 146 |
+
],
|
| 147 |
+
[
|
| 148 |
+
2304,
|
| 149 |
+
1920
|
| 150 |
+
],
|
| 151 |
+
[
|
| 152 |
+
2304,
|
| 153 |
+
2304
|
| 154 |
+
]
|
| 155 |
+
],
|
| 156 |
+
"image_token_index": 151669,
|
| 157 |
+
"model_type": "llava_qwen3_5vit_qwen3",
|
| 158 |
+
"multimodal_projector_bias": true,
|
| 159 |
+
"pad_token_id": 151643,
|
| 160 |
+
"pair_fusion_clip_selection": "",
|
| 161 |
+
"pair_fusion_lr": 0.0001,
|
| 162 |
+
"pair_fusion_motion_cache_format": "",
|
| 163 |
+
"pair_fusion_motion_cache_version": 0,
|
| 164 |
+
"pair_fusion_motion_record_count": 0,
|
| 165 |
+
"pair_fusion_pretraining": "",
|
| 166 |
+
"pair_fusion_pretraining_step": 0,
|
| 167 |
+
"projector_hidden_act": "gelu",
|
| 168 |
+
"shared_image_temporal_refiner": false,
|
| 169 |
+
"temporal_refiner_lr": 0.0001,
|
| 170 |
+
"temporal_refiner_variant": "attention_only",
|
| 171 |
+
"temporal_refiner_window_groups": 4,
|
| 172 |
+
"text_config": {
|
| 173 |
+
"_name_or_path": "/share/m2v_intern_v3/wangjunjie09/model_cache/huggingface/Qwen/Qwen3-8B",
|
| 174 |
+
"architectures": [
|
| 175 |
+
"Qwen3ForCausalLM"
|
| 176 |
+
],
|
| 177 |
+
"attention_bias": false,
|
| 178 |
+
"attention_dropout": 0.0,
|
| 179 |
+
"bos_token_id": 151643,
|
| 180 |
+
"dtype": "bfloat16",
|
| 181 |
+
"eos_token_id": 151645,
|
| 182 |
+
"head_dim": 128,
|
| 183 |
+
"hidden_act": "silu",
|
| 184 |
+
"hidden_size": 4096,
|
| 185 |
+
"initializer_range": 0.02,
|
| 186 |
+
"intermediate_size": 12288,
|
| 187 |
+
"layer_types": [
|
| 188 |
+
"full_attention",
|
| 189 |
+
"full_attention",
|
| 190 |
+
"full_attention",
|
| 191 |
+
"full_attention",
|
| 192 |
+
"full_attention",
|
| 193 |
+
"full_attention",
|
| 194 |
+
"full_attention",
|
| 195 |
+
"full_attention",
|
| 196 |
+
"full_attention",
|
| 197 |
+
"full_attention",
|
| 198 |
+
"full_attention",
|
| 199 |
+
"full_attention",
|
| 200 |
+
"full_attention",
|
| 201 |
+
"full_attention",
|
| 202 |
+
"full_attention",
|
| 203 |
+
"full_attention",
|
| 204 |
+
"full_attention",
|
| 205 |
+
"full_attention",
|
| 206 |
+
"full_attention",
|
| 207 |
+
"full_attention",
|
| 208 |
+
"full_attention",
|
| 209 |
+
"full_attention",
|
| 210 |
+
"full_attention",
|
| 211 |
+
"full_attention",
|
| 212 |
+
"full_attention",
|
| 213 |
+
"full_attention",
|
| 214 |
+
"full_attention",
|
| 215 |
+
"full_attention",
|
| 216 |
+
"full_attention",
|
| 217 |
+
"full_attention",
|
| 218 |
+
"full_attention",
|
| 219 |
+
"full_attention",
|
| 220 |
+
"full_attention",
|
| 221 |
+
"full_attention",
|
| 222 |
+
"full_attention",
|
| 223 |
+
"full_attention"
|
| 224 |
+
],
|
| 225 |
+
"max_position_embeddings": 40960,
|
| 226 |
+
"max_window_layers": 36,
|
| 227 |
+
"model_type": "qwen3",
|
| 228 |
+
"num_attention_heads": 32,
|
| 229 |
+
"num_hidden_layers": 36,
|
| 230 |
+
"num_key_value_heads": 8,
|
| 231 |
+
"pad_token_id": 151643,
|
| 232 |
+
"rms_norm_eps": 1e-06,
|
| 233 |
+
"rope_parameters": {
|
| 234 |
+
"rope_theta": 1000000,
|
| 235 |
+
"rope_type": "default"
|
| 236 |
+
},
|
| 237 |
+
"sliding_window": null,
|
| 238 |
+
"tie_word_embeddings": false,
|
| 239 |
+
"use_cache": false,
|
| 240 |
+
"use_sliding_window": false,
|
| 241 |
+
"vocab_size": 151936
|
| 242 |
+
},
|
| 243 |
+
"tie_word_embeddings": false,
|
| 244 |
+
"train_num_frames": 64,
|
| 245 |
+
"transformers_version": "5.5.4",
|
| 246 |
+
"video_mode": "temporal_merge",
|
| 247 |
+
"video_temporal_refiner": 0,
|
| 248 |
+
"video_token_index": 151670,
|
| 249 |
+
"vision_aspect_ratio": "anyres_max_9",
|
| 250 |
+
"vision_config": {
|
| 251 |
+
"deepstack_visual_indexes": [],
|
| 252 |
+
"depth": 24,
|
| 253 |
+
"dtype": "bfloat16",
|
| 254 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 255 |
+
"hidden_size": 1024,
|
| 256 |
+
"image_size": 384,
|
| 257 |
+
"in_channels": 3,
|
| 258 |
+
"initializer_range": 0.02,
|
| 259 |
+
"intermediate_size": 4096,
|
| 260 |
+
"model_type": "qwen3_5",
|
| 261 |
+
"num_heads": 16,
|
| 262 |
+
"num_position_embeddings": 2304,
|
| 263 |
+
"out_hidden_size": 2048,
|
| 264 |
+
"patch_size": 16,
|
| 265 |
+
"spatial_merge_size": 2,
|
| 266 |
+
"temporal_patch_size": 2
|
| 267 |
+
},
|
| 268 |
+
"vision_feature_layer": -1,
|
| 269 |
+
"vision_feature_select_strategy": "full",
|
| 270 |
+
"vra_layer": 18,
|
| 271 |
+
"vra_loss": 0,
|
| 272 |
+
"vra_scope": "image_video",
|
| 273 |
+
"vra_target_path": "/share/m2v_intern_v3/wangjunjie09/model_cache/huggingface/facebook/dinov3-vitl16-pretrain-lvd1689m",
|
| 274 |
+
"vra_teacher_kind": "dino",
|
| 275 |
+
"vra_teacher_res": 384,
|
| 276 |
+
"vra_video_loss": "direct",
|
| 277 |
+
"vra_weight": 0.5,
|
| 278 |
+
"vra_z_dim": 1024,
|
| 279 |
+
"auto_map": {
|
| 280 |
+
"AutoConfig": "modeling_qwen3_5vit_qwen3.LlavaQwen3_5ViTConfig",
|
| 281 |
+
"AutoModel": "modeling_qwen3_5vit_qwen3.LlavaQwen3_5ViTForConditionalGeneration",
|
| 282 |
+
"AutoModelForCausalLM": "modeling_qwen3_5vit_qwen3.LlavaQwen3_5ViTForConditionalGeneration",
|
| 283 |
+
"AutoModelForImageTextToText": "modeling_qwen3_5vit_qwen3.LlavaQwen3_5ViTForConditionalGeneration"
|
| 284 |
+
}
|
| 285 |
+
}
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/generation_config.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 151643,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
151645,
|
| 6 |
+
151643
|
| 7 |
+
],
|
| 8 |
+
"output_attentions": false,
|
| 9 |
+
"output_hidden_states": false,
|
| 10 |
+
"transformers_version": "5.5.4",
|
| 11 |
+
"use_cache": true
|
| 12 |
+
}
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/model-00001-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:59f325656d3db8a3ad998275879bc608fb71369b180bfa21b29bd48766567aea
|
| 3 |
+
size 4905365464
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/model-00002-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:69ee3fb52e7d6529af2e845a43c76539436dbb137d25261fac9d659852e5162b
|
| 3 |
+
size 4915962456
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/model-00003-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e1d223dcfa1fc59520f282d484bcb86c0d89d43cab0973f397cf4704fa4f53b
|
| 3 |
+
size 4974674312
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/model-00004-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6517c5cd64c78665ad444b0106215346481dad635f286d0e39610bd5d7e5401d
|
| 3 |
+
size 2240014936
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/model.safetensors.index.json
ADDED
|
@@ -0,0 +1,705 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"metadata": {
|
| 3 |
+
"total_parameters": 8517963776,
|
| 4 |
+
"total_size": 17035927552
|
| 5 |
+
},
|
| 6 |
+
"weight_map": {
|
| 7 |
+
"lm_head.weight": "model-00001-of-00004.safetensors",
|
| 8 |
+
"model.image_newline": "model-00001-of-00004.safetensors",
|
| 9 |
+
"model.language_model.embed_tokens.weight": "model-00001-of-00004.safetensors",
|
| 10 |
+
"model.language_model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 11 |
+
"model.language_model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
|
| 12 |
+
"model.language_model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
|
| 13 |
+
"model.language_model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
|
| 14 |
+
"model.language_model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 15 |
+
"model.language_model.layers.0.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 16 |
+
"model.language_model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
|
| 17 |
+
"model.language_model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
|
| 18 |
+
"model.language_model.layers.0.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 19 |
+
"model.language_model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
|
| 20 |
+
"model.language_model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
|
| 21 |
+
"model.language_model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 22 |
+
"model.language_model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
|
| 23 |
+
"model.language_model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
|
| 24 |
+
"model.language_model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
|
| 25 |
+
"model.language_model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 26 |
+
"model.language_model.layers.1.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 27 |
+
"model.language_model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
|
| 28 |
+
"model.language_model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
|
| 29 |
+
"model.language_model.layers.1.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 30 |
+
"model.language_model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
|
| 31 |
+
"model.language_model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
|
| 32 |
+
"model.language_model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 33 |
+
"model.language_model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 34 |
+
"model.language_model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 35 |
+
"model.language_model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 36 |
+
"model.language_model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 37 |
+
"model.language_model.layers.10.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 38 |
+
"model.language_model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 39 |
+
"model.language_model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 40 |
+
"model.language_model.layers.10.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 41 |
+
"model.language_model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 42 |
+
"model.language_model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 43 |
+
"model.language_model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 44 |
+
"model.language_model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 45 |
+
"model.language_model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 46 |
+
"model.language_model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 47 |
+
"model.language_model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 48 |
+
"model.language_model.layers.11.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 49 |
+
"model.language_model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 50 |
+
"model.language_model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 51 |
+
"model.language_model.layers.11.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 52 |
+
"model.language_model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 53 |
+
"model.language_model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 54 |
+
"model.language_model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 55 |
+
"model.language_model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 56 |
+
"model.language_model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 57 |
+
"model.language_model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 58 |
+
"model.language_model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 59 |
+
"model.language_model.layers.12.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 60 |
+
"model.language_model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 61 |
+
"model.language_model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 62 |
+
"model.language_model.layers.12.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 63 |
+
"model.language_model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 64 |
+
"model.language_model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 65 |
+
"model.language_model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 66 |
+
"model.language_model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 67 |
+
"model.language_model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 68 |
+
"model.language_model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 69 |
+
"model.language_model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 70 |
+
"model.language_model.layers.13.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 71 |
+
"model.language_model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 72 |
+
"model.language_model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 73 |
+
"model.language_model.layers.13.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 74 |
+
"model.language_model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 75 |
+
"model.language_model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 76 |
+
"model.language_model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 77 |
+
"model.language_model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 78 |
+
"model.language_model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 79 |
+
"model.language_model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 80 |
+
"model.language_model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 81 |
+
"model.language_model.layers.14.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 82 |
+
"model.language_model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 83 |
+
"model.language_model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 84 |
+
"model.language_model.layers.14.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 85 |
+
"model.language_model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 86 |
+
"model.language_model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 87 |
+
"model.language_model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 88 |
+
"model.language_model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 89 |
+
"model.language_model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 90 |
+
"model.language_model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 91 |
+
"model.language_model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 92 |
+
"model.language_model.layers.15.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 93 |
+
"model.language_model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 94 |
+
"model.language_model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 95 |
+
"model.language_model.layers.15.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 96 |
+
"model.language_model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 97 |
+
"model.language_model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 98 |
+
"model.language_model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 99 |
+
"model.language_model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 100 |
+
"model.language_model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 101 |
+
"model.language_model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 102 |
+
"model.language_model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 103 |
+
"model.language_model.layers.16.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 104 |
+
"model.language_model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 105 |
+
"model.language_model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 106 |
+
"model.language_model.layers.16.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 107 |
+
"model.language_model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 108 |
+
"model.language_model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 109 |
+
"model.language_model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 110 |
+
"model.language_model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 111 |
+
"model.language_model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 112 |
+
"model.language_model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 113 |
+
"model.language_model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 114 |
+
"model.language_model.layers.17.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 115 |
+
"model.language_model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 116 |
+
"model.language_model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 117 |
+
"model.language_model.layers.17.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 118 |
+
"model.language_model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 119 |
+
"model.language_model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 120 |
+
"model.language_model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 121 |
+
"model.language_model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 122 |
+
"model.language_model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 123 |
+
"model.language_model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 124 |
+
"model.language_model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 125 |
+
"model.language_model.layers.18.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 126 |
+
"model.language_model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 127 |
+
"model.language_model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 128 |
+
"model.language_model.layers.18.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 129 |
+
"model.language_model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 130 |
+
"model.language_model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 131 |
+
"model.language_model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 132 |
+
"model.language_model.layers.19.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 133 |
+
"model.language_model.layers.19.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 134 |
+
"model.language_model.layers.19.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 135 |
+
"model.language_model.layers.19.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 136 |
+
"model.language_model.layers.19.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 137 |
+
"model.language_model.layers.19.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 138 |
+
"model.language_model.layers.19.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 139 |
+
"model.language_model.layers.19.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 140 |
+
"model.language_model.layers.19.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 141 |
+
"model.language_model.layers.19.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 142 |
+
"model.language_model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 143 |
+
"model.language_model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
|
| 144 |
+
"model.language_model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
|
| 145 |
+
"model.language_model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
|
| 146 |
+
"model.language_model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 147 |
+
"model.language_model.layers.2.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 148 |
+
"model.language_model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
|
| 149 |
+
"model.language_model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
|
| 150 |
+
"model.language_model.layers.2.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 151 |
+
"model.language_model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
|
| 152 |
+
"model.language_model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
|
| 153 |
+
"model.language_model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 154 |
+
"model.language_model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 155 |
+
"model.language_model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 156 |
+
"model.language_model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 157 |
+
"model.language_model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 158 |
+
"model.language_model.layers.20.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 159 |
+
"model.language_model.layers.20.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 160 |
+
"model.language_model.layers.20.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 161 |
+
"model.language_model.layers.20.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 162 |
+
"model.language_model.layers.20.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 163 |
+
"model.language_model.layers.20.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 164 |
+
"model.language_model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 165 |
+
"model.language_model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 166 |
+
"model.language_model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 167 |
+
"model.language_model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 168 |
+
"model.language_model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 169 |
+
"model.language_model.layers.21.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 170 |
+
"model.language_model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 171 |
+
"model.language_model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 172 |
+
"model.language_model.layers.21.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 173 |
+
"model.language_model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 174 |
+
"model.language_model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 175 |
+
"model.language_model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 176 |
+
"model.language_model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 177 |
+
"model.language_model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 178 |
+
"model.language_model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 179 |
+
"model.language_model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 180 |
+
"model.language_model.layers.22.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 181 |
+
"model.language_model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 182 |
+
"model.language_model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 183 |
+
"model.language_model.layers.22.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 184 |
+
"model.language_model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 185 |
+
"model.language_model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 186 |
+
"model.language_model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 187 |
+
"model.language_model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 188 |
+
"model.language_model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 189 |
+
"model.language_model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 190 |
+
"model.language_model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 191 |
+
"model.language_model.layers.23.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 192 |
+
"model.language_model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 193 |
+
"model.language_model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 194 |
+
"model.language_model.layers.23.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 195 |
+
"model.language_model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 196 |
+
"model.language_model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 197 |
+
"model.language_model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 198 |
+
"model.language_model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 199 |
+
"model.language_model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 200 |
+
"model.language_model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 201 |
+
"model.language_model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 202 |
+
"model.language_model.layers.24.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 203 |
+
"model.language_model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 204 |
+
"model.language_model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 205 |
+
"model.language_model.layers.24.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 206 |
+
"model.language_model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 207 |
+
"model.language_model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 208 |
+
"model.language_model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 209 |
+
"model.language_model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 210 |
+
"model.language_model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 211 |
+
"model.language_model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 212 |
+
"model.language_model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 213 |
+
"model.language_model.layers.25.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 214 |
+
"model.language_model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 215 |
+
"model.language_model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 216 |
+
"model.language_model.layers.25.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 217 |
+
"model.language_model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 218 |
+
"model.language_model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 219 |
+
"model.language_model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 220 |
+
"model.language_model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 221 |
+
"model.language_model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 222 |
+
"model.language_model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 223 |
+
"model.language_model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 224 |
+
"model.language_model.layers.26.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 225 |
+
"model.language_model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 226 |
+
"model.language_model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 227 |
+
"model.language_model.layers.26.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 228 |
+
"model.language_model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 229 |
+
"model.language_model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 230 |
+
"model.language_model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 231 |
+
"model.language_model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 232 |
+
"model.language_model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 233 |
+
"model.language_model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 234 |
+
"model.language_model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 235 |
+
"model.language_model.layers.27.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 236 |
+
"model.language_model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 237 |
+
"model.language_model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 238 |
+
"model.language_model.layers.27.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 239 |
+
"model.language_model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 240 |
+
"model.language_model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 241 |
+
"model.language_model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 242 |
+
"model.language_model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 243 |
+
"model.language_model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 244 |
+
"model.language_model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 245 |
+
"model.language_model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 246 |
+
"model.language_model.layers.28.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 247 |
+
"model.language_model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 248 |
+
"model.language_model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 249 |
+
"model.language_model.layers.28.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 250 |
+
"model.language_model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 251 |
+
"model.language_model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 252 |
+
"model.language_model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 253 |
+
"model.language_model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 254 |
+
"model.language_model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 255 |
+
"model.language_model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 256 |
+
"model.language_model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 257 |
+
"model.language_model.layers.29.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 258 |
+
"model.language_model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 259 |
+
"model.language_model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 260 |
+
"model.language_model.layers.29.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 261 |
+
"model.language_model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 262 |
+
"model.language_model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 263 |
+
"model.language_model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 264 |
+
"model.language_model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
|
| 265 |
+
"model.language_model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
|
| 266 |
+
"model.language_model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
|
| 267 |
+
"model.language_model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 268 |
+
"model.language_model.layers.3.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 269 |
+
"model.language_model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
|
| 270 |
+
"model.language_model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
|
| 271 |
+
"model.language_model.layers.3.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 272 |
+
"model.language_model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
|
| 273 |
+
"model.language_model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
|
| 274 |
+
"model.language_model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 275 |
+
"model.language_model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 276 |
+
"model.language_model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 277 |
+
"model.language_model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 278 |
+
"model.language_model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 279 |
+
"model.language_model.layers.30.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 280 |
+
"model.language_model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 281 |
+
"model.language_model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 282 |
+
"model.language_model.layers.30.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 283 |
+
"model.language_model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 284 |
+
"model.language_model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 285 |
+
"model.language_model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 286 |
+
"model.language_model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 287 |
+
"model.language_model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 288 |
+
"model.language_model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 289 |
+
"model.language_model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 290 |
+
"model.language_model.layers.31.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 291 |
+
"model.language_model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 292 |
+
"model.language_model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 293 |
+
"model.language_model.layers.31.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 294 |
+
"model.language_model.layers.31.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
|
| 295 |
+
"model.language_model.layers.31.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
|
| 296 |
+
"model.language_model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 297 |
+
"model.language_model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
|
| 298 |
+
"model.language_model.layers.32.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
|
| 299 |
+
"model.language_model.layers.32.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
|
| 300 |
+
"model.language_model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 301 |
+
"model.language_model.layers.32.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
|
| 302 |
+
"model.language_model.layers.32.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
|
| 303 |
+
"model.language_model.layers.32.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
|
| 304 |
+
"model.language_model.layers.32.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
|
| 305 |
+
"model.language_model.layers.32.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
|
| 306 |
+
"model.language_model.layers.32.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
|
| 307 |
+
"model.language_model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 308 |
+
"model.language_model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
|
| 309 |
+
"model.language_model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
|
| 310 |
+
"model.language_model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
|
| 311 |
+
"model.language_model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 312 |
+
"model.language_model.layers.33.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
|
| 313 |
+
"model.language_model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
|
| 314 |
+
"model.language_model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
|
| 315 |
+
"model.language_model.layers.33.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
|
| 316 |
+
"model.language_model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
|
| 317 |
+
"model.language_model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
|
| 318 |
+
"model.language_model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 319 |
+
"model.language_model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
|
| 320 |
+
"model.language_model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
|
| 321 |
+
"model.language_model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
|
| 322 |
+
"model.language_model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 323 |
+
"model.language_model.layers.34.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
|
| 324 |
+
"model.language_model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
|
| 325 |
+
"model.language_model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
|
| 326 |
+
"model.language_model.layers.34.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
|
| 327 |
+
"model.language_model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
|
| 328 |
+
"model.language_model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
|
| 329 |
+
"model.language_model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 330 |
+
"model.language_model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
|
| 331 |
+
"model.language_model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
|
| 332 |
+
"model.language_model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
|
| 333 |
+
"model.language_model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 334 |
+
"model.language_model.layers.35.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
|
| 335 |
+
"model.language_model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
|
| 336 |
+
"model.language_model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
|
| 337 |
+
"model.language_model.layers.35.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
|
| 338 |
+
"model.language_model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
|
| 339 |
+
"model.language_model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
|
| 340 |
+
"model.language_model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 341 |
+
"model.language_model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
|
| 342 |
+
"model.language_model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
|
| 343 |
+
"model.language_model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
|
| 344 |
+
"model.language_model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 345 |
+
"model.language_model.layers.4.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 346 |
+
"model.language_model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
|
| 347 |
+
"model.language_model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
|
| 348 |
+
"model.language_model.layers.4.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 349 |
+
"model.language_model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
|
| 350 |
+
"model.language_model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
|
| 351 |
+
"model.language_model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 352 |
+
"model.language_model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
|
| 353 |
+
"model.language_model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
|
| 354 |
+
"model.language_model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
|
| 355 |
+
"model.language_model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 356 |
+
"model.language_model.layers.5.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 357 |
+
"model.language_model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
|
| 358 |
+
"model.language_model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
|
| 359 |
+
"model.language_model.layers.5.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 360 |
+
"model.language_model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
|
| 361 |
+
"model.language_model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
|
| 362 |
+
"model.language_model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 363 |
+
"model.language_model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
|
| 364 |
+
"model.language_model.layers.6.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 365 |
+
"model.language_model.layers.6.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 366 |
+
"model.language_model.layers.6.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 367 |
+
"model.language_model.layers.6.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 368 |
+
"model.language_model.layers.6.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 369 |
+
"model.language_model.layers.6.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 370 |
+
"model.language_model.layers.6.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 371 |
+
"model.language_model.layers.6.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 372 |
+
"model.language_model.layers.6.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 373 |
+
"model.language_model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 374 |
+
"model.language_model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 375 |
+
"model.language_model.layers.7.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 376 |
+
"model.language_model.layers.7.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 377 |
+
"model.language_model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 378 |
+
"model.language_model.layers.7.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 379 |
+
"model.language_model.layers.7.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 380 |
+
"model.language_model.layers.7.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 381 |
+
"model.language_model.layers.7.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 382 |
+
"model.language_model.layers.7.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 383 |
+
"model.language_model.layers.7.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 384 |
+
"model.language_model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 385 |
+
"model.language_model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 386 |
+
"model.language_model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 387 |
+
"model.language_model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 388 |
+
"model.language_model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 389 |
+
"model.language_model.layers.8.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 390 |
+
"model.language_model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 391 |
+
"model.language_model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 392 |
+
"model.language_model.layers.8.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 393 |
+
"model.language_model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 394 |
+
"model.language_model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 395 |
+
"model.language_model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 396 |
+
"model.language_model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 397 |
+
"model.language_model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 398 |
+
"model.language_model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 399 |
+
"model.language_model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 400 |
+
"model.language_model.layers.9.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 401 |
+
"model.language_model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 402 |
+
"model.language_model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 403 |
+
"model.language_model.layers.9.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 404 |
+
"model.language_model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 405 |
+
"model.language_model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 406 |
+
"model.language_model.norm.weight": "model-00004-of-00004.safetensors",
|
| 407 |
+
"model.multi_modal_projector.linear_1.bias": "model-00004-of-00004.safetensors",
|
| 408 |
+
"model.multi_modal_projector.linear_1.weight": "model-00004-of-00004.safetensors",
|
| 409 |
+
"model.multi_modal_projector.linear_2.bias": "model-00004-of-00004.safetensors",
|
| 410 |
+
"model.multi_modal_projector.linear_2.weight": "model-00004-of-00004.safetensors",
|
| 411 |
+
"model.vision_tower.vision.blocks.0.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 412 |
+
"model.vision_tower.vision.blocks.0.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 413 |
+
"model.vision_tower.vision.blocks.0.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 414 |
+
"model.vision_tower.vision.blocks.0.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 415 |
+
"model.vision_tower.vision.blocks.0.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 416 |
+
"model.vision_tower.vision.blocks.0.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 417 |
+
"model.vision_tower.vision.blocks.0.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 418 |
+
"model.vision_tower.vision.blocks.0.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 419 |
+
"model.vision_tower.vision.blocks.0.norm1.bias": "model-00004-of-00004.safetensors",
|
| 420 |
+
"model.vision_tower.vision.blocks.0.norm1.weight": "model-00004-of-00004.safetensors",
|
| 421 |
+
"model.vision_tower.vision.blocks.0.norm2.bias": "model-00004-of-00004.safetensors",
|
| 422 |
+
"model.vision_tower.vision.blocks.0.norm2.weight": "model-00004-of-00004.safetensors",
|
| 423 |
+
"model.vision_tower.vision.blocks.1.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 424 |
+
"model.vision_tower.vision.blocks.1.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 425 |
+
"model.vision_tower.vision.blocks.1.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 426 |
+
"model.vision_tower.vision.blocks.1.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 427 |
+
"model.vision_tower.vision.blocks.1.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 428 |
+
"model.vision_tower.vision.blocks.1.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 429 |
+
"model.vision_tower.vision.blocks.1.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 430 |
+
"model.vision_tower.vision.blocks.1.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 431 |
+
"model.vision_tower.vision.blocks.1.norm1.bias": "model-00004-of-00004.safetensors",
|
| 432 |
+
"model.vision_tower.vision.blocks.1.norm1.weight": "model-00004-of-00004.safetensors",
|
| 433 |
+
"model.vision_tower.vision.blocks.1.norm2.bias": "model-00004-of-00004.safetensors",
|
| 434 |
+
"model.vision_tower.vision.blocks.1.norm2.weight": "model-00004-of-00004.safetensors",
|
| 435 |
+
"model.vision_tower.vision.blocks.10.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 436 |
+
"model.vision_tower.vision.blocks.10.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 437 |
+
"model.vision_tower.vision.blocks.10.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 438 |
+
"model.vision_tower.vision.blocks.10.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 439 |
+
"model.vision_tower.vision.blocks.10.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 440 |
+
"model.vision_tower.vision.blocks.10.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 441 |
+
"model.vision_tower.vision.blocks.10.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 442 |
+
"model.vision_tower.vision.blocks.10.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 443 |
+
"model.vision_tower.vision.blocks.10.norm1.bias": "model-00004-of-00004.safetensors",
|
| 444 |
+
"model.vision_tower.vision.blocks.10.norm1.weight": "model-00004-of-00004.safetensors",
|
| 445 |
+
"model.vision_tower.vision.blocks.10.norm2.bias": "model-00004-of-00004.safetensors",
|
| 446 |
+
"model.vision_tower.vision.blocks.10.norm2.weight": "model-00004-of-00004.safetensors",
|
| 447 |
+
"model.vision_tower.vision.blocks.11.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 448 |
+
"model.vision_tower.vision.blocks.11.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 449 |
+
"model.vision_tower.vision.blocks.11.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 450 |
+
"model.vision_tower.vision.blocks.11.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 451 |
+
"model.vision_tower.vision.blocks.11.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 452 |
+
"model.vision_tower.vision.blocks.11.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 453 |
+
"model.vision_tower.vision.blocks.11.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 454 |
+
"model.vision_tower.vision.blocks.11.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 455 |
+
"model.vision_tower.vision.blocks.11.norm1.bias": "model-00004-of-00004.safetensors",
|
| 456 |
+
"model.vision_tower.vision.blocks.11.norm1.weight": "model-00004-of-00004.safetensors",
|
| 457 |
+
"model.vision_tower.vision.blocks.11.norm2.bias": "model-00004-of-00004.safetensors",
|
| 458 |
+
"model.vision_tower.vision.blocks.11.norm2.weight": "model-00004-of-00004.safetensors",
|
| 459 |
+
"model.vision_tower.vision.blocks.12.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 460 |
+
"model.vision_tower.vision.blocks.12.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 461 |
+
"model.vision_tower.vision.blocks.12.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 462 |
+
"model.vision_tower.vision.blocks.12.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 463 |
+
"model.vision_tower.vision.blocks.12.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 464 |
+
"model.vision_tower.vision.blocks.12.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 465 |
+
"model.vision_tower.vision.blocks.12.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 466 |
+
"model.vision_tower.vision.blocks.12.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 467 |
+
"model.vision_tower.vision.blocks.12.norm1.bias": "model-00004-of-00004.safetensors",
|
| 468 |
+
"model.vision_tower.vision.blocks.12.norm1.weight": "model-00004-of-00004.safetensors",
|
| 469 |
+
"model.vision_tower.vision.blocks.12.norm2.bias": "model-00004-of-00004.safetensors",
|
| 470 |
+
"model.vision_tower.vision.blocks.12.norm2.weight": "model-00004-of-00004.safetensors",
|
| 471 |
+
"model.vision_tower.vision.blocks.13.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 472 |
+
"model.vision_tower.vision.blocks.13.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 473 |
+
"model.vision_tower.vision.blocks.13.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 474 |
+
"model.vision_tower.vision.blocks.13.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 475 |
+
"model.vision_tower.vision.blocks.13.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 476 |
+
"model.vision_tower.vision.blocks.13.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 477 |
+
"model.vision_tower.vision.blocks.13.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 478 |
+
"model.vision_tower.vision.blocks.13.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 479 |
+
"model.vision_tower.vision.blocks.13.norm1.bias": "model-00004-of-00004.safetensors",
|
| 480 |
+
"model.vision_tower.vision.blocks.13.norm1.weight": "model-00004-of-00004.safetensors",
|
| 481 |
+
"model.vision_tower.vision.blocks.13.norm2.bias": "model-00004-of-00004.safetensors",
|
| 482 |
+
"model.vision_tower.vision.blocks.13.norm2.weight": "model-00004-of-00004.safetensors",
|
| 483 |
+
"model.vision_tower.vision.blocks.14.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 484 |
+
"model.vision_tower.vision.blocks.14.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 485 |
+
"model.vision_tower.vision.blocks.14.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 486 |
+
"model.vision_tower.vision.blocks.14.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 487 |
+
"model.vision_tower.vision.blocks.14.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 488 |
+
"model.vision_tower.vision.blocks.14.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 489 |
+
"model.vision_tower.vision.blocks.14.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 490 |
+
"model.vision_tower.vision.blocks.14.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 491 |
+
"model.vision_tower.vision.blocks.14.norm1.bias": "model-00004-of-00004.safetensors",
|
| 492 |
+
"model.vision_tower.vision.blocks.14.norm1.weight": "model-00004-of-00004.safetensors",
|
| 493 |
+
"model.vision_tower.vision.blocks.14.norm2.bias": "model-00004-of-00004.safetensors",
|
| 494 |
+
"model.vision_tower.vision.blocks.14.norm2.weight": "model-00004-of-00004.safetensors",
|
| 495 |
+
"model.vision_tower.vision.blocks.15.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 496 |
+
"model.vision_tower.vision.blocks.15.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 497 |
+
"model.vision_tower.vision.blocks.15.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 498 |
+
"model.vision_tower.vision.blocks.15.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 499 |
+
"model.vision_tower.vision.blocks.15.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 500 |
+
"model.vision_tower.vision.blocks.15.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 501 |
+
"model.vision_tower.vision.blocks.15.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 502 |
+
"model.vision_tower.vision.blocks.15.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 503 |
+
"model.vision_tower.vision.blocks.15.norm1.bias": "model-00004-of-00004.safetensors",
|
| 504 |
+
"model.vision_tower.vision.blocks.15.norm1.weight": "model-00004-of-00004.safetensors",
|
| 505 |
+
"model.vision_tower.vision.blocks.15.norm2.bias": "model-00004-of-00004.safetensors",
|
| 506 |
+
"model.vision_tower.vision.blocks.15.norm2.weight": "model-00004-of-00004.safetensors",
|
| 507 |
+
"model.vision_tower.vision.blocks.16.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 508 |
+
"model.vision_tower.vision.blocks.16.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 509 |
+
"model.vision_tower.vision.blocks.16.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 510 |
+
"model.vision_tower.vision.blocks.16.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 511 |
+
"model.vision_tower.vision.blocks.16.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 512 |
+
"model.vision_tower.vision.blocks.16.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 513 |
+
"model.vision_tower.vision.blocks.16.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 514 |
+
"model.vision_tower.vision.blocks.16.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 515 |
+
"model.vision_tower.vision.blocks.16.norm1.bias": "model-00004-of-00004.safetensors",
|
| 516 |
+
"model.vision_tower.vision.blocks.16.norm1.weight": "model-00004-of-00004.safetensors",
|
| 517 |
+
"model.vision_tower.vision.blocks.16.norm2.bias": "model-00004-of-00004.safetensors",
|
| 518 |
+
"model.vision_tower.vision.blocks.16.norm2.weight": "model-00004-of-00004.safetensors",
|
| 519 |
+
"model.vision_tower.vision.blocks.17.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 520 |
+
"model.vision_tower.vision.blocks.17.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 521 |
+
"model.vision_tower.vision.blocks.17.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 522 |
+
"model.vision_tower.vision.blocks.17.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 523 |
+
"model.vision_tower.vision.blocks.17.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 524 |
+
"model.vision_tower.vision.blocks.17.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 525 |
+
"model.vision_tower.vision.blocks.17.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 526 |
+
"model.vision_tower.vision.blocks.17.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 527 |
+
"model.vision_tower.vision.blocks.17.norm1.bias": "model-00004-of-00004.safetensors",
|
| 528 |
+
"model.vision_tower.vision.blocks.17.norm1.weight": "model-00004-of-00004.safetensors",
|
| 529 |
+
"model.vision_tower.vision.blocks.17.norm2.bias": "model-00004-of-00004.safetensors",
|
| 530 |
+
"model.vision_tower.vision.blocks.17.norm2.weight": "model-00004-of-00004.safetensors",
|
| 531 |
+
"model.vision_tower.vision.blocks.18.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 532 |
+
"model.vision_tower.vision.blocks.18.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 533 |
+
"model.vision_tower.vision.blocks.18.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 534 |
+
"model.vision_tower.vision.blocks.18.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 535 |
+
"model.vision_tower.vision.blocks.18.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 536 |
+
"model.vision_tower.vision.blocks.18.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 537 |
+
"model.vision_tower.vision.blocks.18.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 538 |
+
"model.vision_tower.vision.blocks.18.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 539 |
+
"model.vision_tower.vision.blocks.18.norm1.bias": "model-00004-of-00004.safetensors",
|
| 540 |
+
"model.vision_tower.vision.blocks.18.norm1.weight": "model-00004-of-00004.safetensors",
|
| 541 |
+
"model.vision_tower.vision.blocks.18.norm2.bias": "model-00004-of-00004.safetensors",
|
| 542 |
+
"model.vision_tower.vision.blocks.18.norm2.weight": "model-00004-of-00004.safetensors",
|
| 543 |
+
"model.vision_tower.vision.blocks.19.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 544 |
+
"model.vision_tower.vision.blocks.19.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 545 |
+
"model.vision_tower.vision.blocks.19.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 546 |
+
"model.vision_tower.vision.blocks.19.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 547 |
+
"model.vision_tower.vision.blocks.19.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 548 |
+
"model.vision_tower.vision.blocks.19.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 549 |
+
"model.vision_tower.vision.blocks.19.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 550 |
+
"model.vision_tower.vision.blocks.19.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 551 |
+
"model.vision_tower.vision.blocks.19.norm1.bias": "model-00004-of-00004.safetensors",
|
| 552 |
+
"model.vision_tower.vision.blocks.19.norm1.weight": "model-00004-of-00004.safetensors",
|
| 553 |
+
"model.vision_tower.vision.blocks.19.norm2.bias": "model-00004-of-00004.safetensors",
|
| 554 |
+
"model.vision_tower.vision.blocks.19.norm2.weight": "model-00004-of-00004.safetensors",
|
| 555 |
+
"model.vision_tower.vision.blocks.2.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 556 |
+
"model.vision_tower.vision.blocks.2.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 557 |
+
"model.vision_tower.vision.blocks.2.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 558 |
+
"model.vision_tower.vision.blocks.2.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 559 |
+
"model.vision_tower.vision.blocks.2.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 560 |
+
"model.vision_tower.vision.blocks.2.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 561 |
+
"model.vision_tower.vision.blocks.2.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 562 |
+
"model.vision_tower.vision.blocks.2.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 563 |
+
"model.vision_tower.vision.blocks.2.norm1.bias": "model-00004-of-00004.safetensors",
|
| 564 |
+
"model.vision_tower.vision.blocks.2.norm1.weight": "model-00004-of-00004.safetensors",
|
| 565 |
+
"model.vision_tower.vision.blocks.2.norm2.bias": "model-00004-of-00004.safetensors",
|
| 566 |
+
"model.vision_tower.vision.blocks.2.norm2.weight": "model-00004-of-00004.safetensors",
|
| 567 |
+
"model.vision_tower.vision.blocks.20.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 568 |
+
"model.vision_tower.vision.blocks.20.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 569 |
+
"model.vision_tower.vision.blocks.20.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 570 |
+
"model.vision_tower.vision.blocks.20.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 571 |
+
"model.vision_tower.vision.blocks.20.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 572 |
+
"model.vision_tower.vision.blocks.20.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 573 |
+
"model.vision_tower.vision.blocks.20.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 574 |
+
"model.vision_tower.vision.blocks.20.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 575 |
+
"model.vision_tower.vision.blocks.20.norm1.bias": "model-00004-of-00004.safetensors",
|
| 576 |
+
"model.vision_tower.vision.blocks.20.norm1.weight": "model-00004-of-00004.safetensors",
|
| 577 |
+
"model.vision_tower.vision.blocks.20.norm2.bias": "model-00004-of-00004.safetensors",
|
| 578 |
+
"model.vision_tower.vision.blocks.20.norm2.weight": "model-00004-of-00004.safetensors",
|
| 579 |
+
"model.vision_tower.vision.blocks.21.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 580 |
+
"model.vision_tower.vision.blocks.21.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 581 |
+
"model.vision_tower.vision.blocks.21.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 582 |
+
"model.vision_tower.vision.blocks.21.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 583 |
+
"model.vision_tower.vision.blocks.21.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 584 |
+
"model.vision_tower.vision.blocks.21.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 585 |
+
"model.vision_tower.vision.blocks.21.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 586 |
+
"model.vision_tower.vision.blocks.21.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 587 |
+
"model.vision_tower.vision.blocks.21.norm1.bias": "model-00004-of-00004.safetensors",
|
| 588 |
+
"model.vision_tower.vision.blocks.21.norm1.weight": "model-00004-of-00004.safetensors",
|
| 589 |
+
"model.vision_tower.vision.blocks.21.norm2.bias": "model-00004-of-00004.safetensors",
|
| 590 |
+
"model.vision_tower.vision.blocks.21.norm2.weight": "model-00004-of-00004.safetensors",
|
| 591 |
+
"model.vision_tower.vision.blocks.22.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 592 |
+
"model.vision_tower.vision.blocks.22.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 593 |
+
"model.vision_tower.vision.blocks.22.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 594 |
+
"model.vision_tower.vision.blocks.22.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 595 |
+
"model.vision_tower.vision.blocks.22.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 596 |
+
"model.vision_tower.vision.blocks.22.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 597 |
+
"model.vision_tower.vision.blocks.22.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 598 |
+
"model.vision_tower.vision.blocks.22.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 599 |
+
"model.vision_tower.vision.blocks.22.norm1.bias": "model-00004-of-00004.safetensors",
|
| 600 |
+
"model.vision_tower.vision.blocks.22.norm1.weight": "model-00004-of-00004.safetensors",
|
| 601 |
+
"model.vision_tower.vision.blocks.22.norm2.bias": "model-00004-of-00004.safetensors",
|
| 602 |
+
"model.vision_tower.vision.blocks.22.norm2.weight": "model-00004-of-00004.safetensors",
|
| 603 |
+
"model.vision_tower.vision.blocks.23.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 604 |
+
"model.vision_tower.vision.blocks.23.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 605 |
+
"model.vision_tower.vision.blocks.23.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 606 |
+
"model.vision_tower.vision.blocks.23.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 607 |
+
"model.vision_tower.vision.blocks.23.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 608 |
+
"model.vision_tower.vision.blocks.23.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 609 |
+
"model.vision_tower.vision.blocks.23.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 610 |
+
"model.vision_tower.vision.blocks.23.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 611 |
+
"model.vision_tower.vision.blocks.23.norm1.bias": "model-00004-of-00004.safetensors",
|
| 612 |
+
"model.vision_tower.vision.blocks.23.norm1.weight": "model-00004-of-00004.safetensors",
|
| 613 |
+
"model.vision_tower.vision.blocks.23.norm2.bias": "model-00004-of-00004.safetensors",
|
| 614 |
+
"model.vision_tower.vision.blocks.23.norm2.weight": "model-00004-of-00004.safetensors",
|
| 615 |
+
"model.vision_tower.vision.blocks.3.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 616 |
+
"model.vision_tower.vision.blocks.3.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 617 |
+
"model.vision_tower.vision.blocks.3.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 618 |
+
"model.vision_tower.vision.blocks.3.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 619 |
+
"model.vision_tower.vision.blocks.3.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 620 |
+
"model.vision_tower.vision.blocks.3.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 621 |
+
"model.vision_tower.vision.blocks.3.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 622 |
+
"model.vision_tower.vision.blocks.3.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 623 |
+
"model.vision_tower.vision.blocks.3.norm1.bias": "model-00004-of-00004.safetensors",
|
| 624 |
+
"model.vision_tower.vision.blocks.3.norm1.weight": "model-00004-of-00004.safetensors",
|
| 625 |
+
"model.vision_tower.vision.blocks.3.norm2.bias": "model-00004-of-00004.safetensors",
|
| 626 |
+
"model.vision_tower.vision.blocks.3.norm2.weight": "model-00004-of-00004.safetensors",
|
| 627 |
+
"model.vision_tower.vision.blocks.4.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 628 |
+
"model.vision_tower.vision.blocks.4.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 629 |
+
"model.vision_tower.vision.blocks.4.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 630 |
+
"model.vision_tower.vision.blocks.4.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 631 |
+
"model.vision_tower.vision.blocks.4.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 632 |
+
"model.vision_tower.vision.blocks.4.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 633 |
+
"model.vision_tower.vision.blocks.4.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 634 |
+
"model.vision_tower.vision.blocks.4.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 635 |
+
"model.vision_tower.vision.blocks.4.norm1.bias": "model-00004-of-00004.safetensors",
|
| 636 |
+
"model.vision_tower.vision.blocks.4.norm1.weight": "model-00004-of-00004.safetensors",
|
| 637 |
+
"model.vision_tower.vision.blocks.4.norm2.bias": "model-00004-of-00004.safetensors",
|
| 638 |
+
"model.vision_tower.vision.blocks.4.norm2.weight": "model-00004-of-00004.safetensors",
|
| 639 |
+
"model.vision_tower.vision.blocks.5.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 640 |
+
"model.vision_tower.vision.blocks.5.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 641 |
+
"model.vision_tower.vision.blocks.5.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 642 |
+
"model.vision_tower.vision.blocks.5.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 643 |
+
"model.vision_tower.vision.blocks.5.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 644 |
+
"model.vision_tower.vision.blocks.5.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 645 |
+
"model.vision_tower.vision.blocks.5.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 646 |
+
"model.vision_tower.vision.blocks.5.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 647 |
+
"model.vision_tower.vision.blocks.5.norm1.bias": "model-00004-of-00004.safetensors",
|
| 648 |
+
"model.vision_tower.vision.blocks.5.norm1.weight": "model-00004-of-00004.safetensors",
|
| 649 |
+
"model.vision_tower.vision.blocks.5.norm2.bias": "model-00004-of-00004.safetensors",
|
| 650 |
+
"model.vision_tower.vision.blocks.5.norm2.weight": "model-00004-of-00004.safetensors",
|
| 651 |
+
"model.vision_tower.vision.blocks.6.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 652 |
+
"model.vision_tower.vision.blocks.6.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 653 |
+
"model.vision_tower.vision.blocks.6.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 654 |
+
"model.vision_tower.vision.blocks.6.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 655 |
+
"model.vision_tower.vision.blocks.6.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 656 |
+
"model.vision_tower.vision.blocks.6.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 657 |
+
"model.vision_tower.vision.blocks.6.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 658 |
+
"model.vision_tower.vision.blocks.6.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 659 |
+
"model.vision_tower.vision.blocks.6.norm1.bias": "model-00004-of-00004.safetensors",
|
| 660 |
+
"model.vision_tower.vision.blocks.6.norm1.weight": "model-00004-of-00004.safetensors",
|
| 661 |
+
"model.vision_tower.vision.blocks.6.norm2.bias": "model-00004-of-00004.safetensors",
|
| 662 |
+
"model.vision_tower.vision.blocks.6.norm2.weight": "model-00004-of-00004.safetensors",
|
| 663 |
+
"model.vision_tower.vision.blocks.7.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 664 |
+
"model.vision_tower.vision.blocks.7.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 665 |
+
"model.vision_tower.vision.blocks.7.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 666 |
+
"model.vision_tower.vision.blocks.7.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 667 |
+
"model.vision_tower.vision.blocks.7.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 668 |
+
"model.vision_tower.vision.blocks.7.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 669 |
+
"model.vision_tower.vision.blocks.7.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 670 |
+
"model.vision_tower.vision.blocks.7.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 671 |
+
"model.vision_tower.vision.blocks.7.norm1.bias": "model-00004-of-00004.safetensors",
|
| 672 |
+
"model.vision_tower.vision.blocks.7.norm1.weight": "model-00004-of-00004.safetensors",
|
| 673 |
+
"model.vision_tower.vision.blocks.7.norm2.bias": "model-00004-of-00004.safetensors",
|
| 674 |
+
"model.vision_tower.vision.blocks.7.norm2.weight": "model-00004-of-00004.safetensors",
|
| 675 |
+
"model.vision_tower.vision.blocks.8.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 676 |
+
"model.vision_tower.vision.blocks.8.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 677 |
+
"model.vision_tower.vision.blocks.8.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 678 |
+
"model.vision_tower.vision.blocks.8.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 679 |
+
"model.vision_tower.vision.blocks.8.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 680 |
+
"model.vision_tower.vision.blocks.8.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 681 |
+
"model.vision_tower.vision.blocks.8.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 682 |
+
"model.vision_tower.vision.blocks.8.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 683 |
+
"model.vision_tower.vision.blocks.8.norm1.bias": "model-00004-of-00004.safetensors",
|
| 684 |
+
"model.vision_tower.vision.blocks.8.norm1.weight": "model-00004-of-00004.safetensors",
|
| 685 |
+
"model.vision_tower.vision.blocks.8.norm2.bias": "model-00004-of-00004.safetensors",
|
| 686 |
+
"model.vision_tower.vision.blocks.8.norm2.weight": "model-00004-of-00004.safetensors",
|
| 687 |
+
"model.vision_tower.vision.blocks.9.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 688 |
+
"model.vision_tower.vision.blocks.9.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 689 |
+
"model.vision_tower.vision.blocks.9.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 690 |
+
"model.vision_tower.vision.blocks.9.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 691 |
+
"model.vision_tower.vision.blocks.9.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 692 |
+
"model.vision_tower.vision.blocks.9.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 693 |
+
"model.vision_tower.vision.blocks.9.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 694 |
+
"model.vision_tower.vision.blocks.9.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 695 |
+
"model.vision_tower.vision.blocks.9.norm1.bias": "model-00004-of-00004.safetensors",
|
| 696 |
+
"model.vision_tower.vision.blocks.9.norm1.weight": "model-00004-of-00004.safetensors",
|
| 697 |
+
"model.vision_tower.vision.blocks.9.norm2.bias": "model-00004-of-00004.safetensors",
|
| 698 |
+
"model.vision_tower.vision.blocks.9.norm2.weight": "model-00004-of-00004.safetensors",
|
| 699 |
+
"model.vision_tower.vision.final_layernorm.bias": "model-00004-of-00004.safetensors",
|
| 700 |
+
"model.vision_tower.vision.final_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 701 |
+
"model.vision_tower.vision.patch_embed.proj.bias": "model-00004-of-00004.safetensors",
|
| 702 |
+
"model.vision_tower.vision.patch_embed.proj.weight": "model-00004-of-00004.safetensors",
|
| 703 |
+
"model.vision_tower.vision.pos_embed.weight": "model-00004-of-00004.safetensors"
|
| 704 |
+
}
|
| 705 |
+
}
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/modeling_qwen3_5vit_qwen3.py
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/processing_qwen3_5vit_qwen3.py
ADDED
|
@@ -0,0 +1,302 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""processing_qwen3_5vit_qwen3.py — LlavaOnevision processor with optional
|
| 2 |
+
video temporal-merge (tmerge) token accounting.
|
| 3 |
+
|
| 4 |
+
Why a subclass: at eval the model is loaded via HF (lmms-eval `llava_hf` →
|
| 5 |
+
`AutoProcessor.from_pretrained(ckpt, trust_remote_code=True)`), NOT the swift
|
| 6 |
+
template. The `LlavaOnevisionProcessor.__call__` expands `<video>` by
|
| 7 |
+
`num_frames*144+1`. Under `video_temporal_merge=1` the model emits
|
| 8 |
+
`(num_frames//2)*144+1` features (grid_t=T/2 temporal groups), so the default
|
| 9 |
+
count over-expands → `masked_scatter` token/feature mismatch crash.
|
| 10 |
+
|
| 11 |
+
This subclass is byte-identical to the upstream processor EXCEPT the single video
|
| 12 |
+
token-count line, which routes through `qwen3_5_video_token_count` — the same
|
| 13 |
+
SoT helper the swift training template uses. The mllm_eval adapter
|
| 14 |
+
(`mllm_eval/checkpoint/__init__.py::qwen3_5vit`) declares this file as
|
| 15 |
+
`processor_file`, so prep cp's it into the ckpt and injects `auto_map.AutoProcessor`.
|
| 16 |
+
|
| 17 |
+
`video_temporal_merge` is a persisted init kwarg (lands in processor_config.json),
|
| 18 |
+
set by the plugin `get_processor` from env/config so eval inherits it with no env.
|
| 19 |
+
"""
|
| 20 |
+
|
| 21 |
+
import math
|
| 22 |
+
|
| 23 |
+
import numpy as np
|
| 24 |
+
|
| 25 |
+
from transformers.feature_extraction_utils import BatchFeature
|
| 26 |
+
from transformers.image_processing_utils import select_best_resolution
|
| 27 |
+
from transformers.image_utils import ImageInput, get_image_size, to_numpy_array
|
| 28 |
+
from transformers.processing_utils import Unpack
|
| 29 |
+
from transformers.tokenization_utils_base import PreTokenizedInput, TextInput
|
| 30 |
+
from transformers.video_utils import VideoInput
|
| 31 |
+
from transformers.models.llava_onevision.processing_llava_onevision import (
|
| 32 |
+
LlavaOnevisionProcessor,
|
| 33 |
+
LlavaOnevisionProcessorKwargs,
|
| 34 |
+
)
|
| 35 |
+
|
| 36 |
+
from typing import Optional
|
| 37 |
+
|
| 38 |
+
# Flat-name sibling import: works both in-repo (model_plugin adds video_mllm/model/ to
|
| 39 |
+
# sys.path) and at eval (HF dynamic_module loads modeling_qwen3_5vit_qwen3.py as a
|
| 40 |
+
# ckpt-dir sibling — the qwen3_5vit adapter ships it as modeling_file).
|
| 41 |
+
from modeling_qwen3_5vit_qwen3 import (
|
| 42 |
+
build_faithful_video_token_string,
|
| 43 |
+
vtc_compression_factor,
|
| 44 |
+
vtc_temporal_patch_size,
|
| 45 |
+
qwen3_5_video_token_count,
|
| 46 |
+
video_is_vtc_compression,
|
| 47 |
+
video_is_spatial_catp,
|
| 48 |
+
video_is_image_mrope,
|
| 49 |
+
video_is_pair_fusion,
|
| 50 |
+
video_is_temporal_compression,
|
| 51 |
+
video_is_temporal_merge,
|
| 52 |
+
video_mrope_active,
|
| 53 |
+
)
|
| 54 |
+
|
| 55 |
+
|
| 56 |
+
class LlavaQwen3_5ViTProcessor(LlavaOnevisionProcessor):
|
| 57 |
+
"""LlavaOnevisionProcessor whose `<video>` expansion is tmerge-aware."""
|
| 58 |
+
|
| 59 |
+
def __init__(
|
| 60 |
+
self,
|
| 61 |
+
image_processor=None,
|
| 62 |
+
tokenizer=None,
|
| 63 |
+
video_processor=None,
|
| 64 |
+
num_image_tokens=None,
|
| 65 |
+
vision_feature_select_strategy=None,
|
| 66 |
+
chat_template=None,
|
| 67 |
+
image_token="<image>",
|
| 68 |
+
video_token="<video>",
|
| 69 |
+
vision_aspect_ratio="anyres_max_9",
|
| 70 |
+
video_mode: Optional[str] = None,
|
| 71 |
+
vtc_compression_factor: int = 2,
|
| 72 |
+
vtc_temporal_patch_size: int = 1,
|
| 73 |
+
# Legacy int flags kept in the signature ONLY so a pre-2026-06-22 processor_config.json
|
| 74 |
+
# still reaches __init__ on reload (ProcessorMixin filters keys NOT in the signature BEFORE
|
| 75 |
+
# calling __init__). Folded into video_mode and NEVER set as attributes -> to_dict (which
|
| 76 |
+
# dumps __dict__) serializes ONLY video_mode (single SoT; old ckpts migrate on re-save).
|
| 77 |
+
video_temporal_merge: int = 0,
|
| 78 |
+
image_mrope: int = 0,
|
| 79 |
+
video_mrope: int = 0,
|
| 80 |
+
**kwargs,
|
| 81 |
+
):
|
| 82 |
+
# The parent's sub-processors AND scalar kwargs MUST be explicit in THIS signature -- NOT
|
| 83 |
+
# swallowed by a bare (*args, **kwargs). ProcessorMixin serialization enumerates what to
|
| 84 |
+
# persist via get_attributes() = inspect.signature(cls.__init__).parameters; a (*args,
|
| 85 |
+
# **kwargs) signature returns [] -> save_pretrained writes neither the sub-processor files
|
| 86 |
+
# nor video_mode into the ckpt. So video_mode is an explicit, serialized param here.
|
| 87 |
+
# Backward-compat: an old processor_config.json carries the legacy int flags (received via
|
| 88 |
+
# the explicit params above so reload reaches here) and no video_mode. Fold them into
|
| 89 |
+
# video_mode; do NOT set them as attributes, so to_dict writes only video_mode (single SoT;
|
| 90 |
+
# old ckpts migrate on re-save). New ckpts pass video_mode directly (fold skipped).
|
| 91 |
+
if video_mode is None:
|
| 92 |
+
if int(image_mrope or 0):
|
| 93 |
+
video_mode = "image_mrope"
|
| 94 |
+
elif int(video_mrope or 0):
|
| 95 |
+
video_mode = "video_mrope"
|
| 96 |
+
elif int(video_temporal_merge or 0):
|
| 97 |
+
video_mode = "temporal_merge"
|
| 98 |
+
self.video_mode = video_mode
|
| 99 |
+
self.vtc_compression_factor = vtc_compression_factor
|
| 100 |
+
self.vtc_temporal_patch_size = vtc_temporal_patch_size
|
| 101 |
+
super().__init__(
|
| 102 |
+
image_processor=image_processor,
|
| 103 |
+
tokenizer=tokenizer,
|
| 104 |
+
video_processor=video_processor,
|
| 105 |
+
num_image_tokens=num_image_tokens,
|
| 106 |
+
vision_feature_select_strategy=vision_feature_select_strategy,
|
| 107 |
+
chat_template=chat_template,
|
| 108 |
+
image_token=image_token,
|
| 109 |
+
video_token=video_token,
|
| 110 |
+
vision_aspect_ratio=vision_aspect_ratio,
|
| 111 |
+
**kwargs,
|
| 112 |
+
)
|
| 113 |
+
|
| 114 |
+
def __call__(
|
| 115 |
+
self,
|
| 116 |
+
images: ImageInput = None,
|
| 117 |
+
text: "TextInput | PreTokenizedInput | list[TextInput] | list[PreTokenizedInput]" = None,
|
| 118 |
+
videos: VideoInput = None,
|
| 119 |
+
video_fps: float = None,
|
| 120 |
+
video_sample_indices=None,
|
| 121 |
+
**kwargs: Unpack[LlavaOnevisionProcessorKwargs],
|
| 122 |
+
) -> BatchFeature:
|
| 123 |
+
# ── verbatim from LlavaOnevisionProcessor.__call__ (transformers) ──
|
| 124 |
+
output_kwargs = self._merge_kwargs(
|
| 125 |
+
LlavaOnevisionProcessorKwargs,
|
| 126 |
+
tokenizer_init_kwargs=self.tokenizer.init_kwargs,
|
| 127 |
+
**kwargs,
|
| 128 |
+
)
|
| 129 |
+
|
| 130 |
+
if isinstance(text, str):
|
| 131 |
+
text = [text]
|
| 132 |
+
elif not isinstance(text, list) and not isinstance(text[0], str):
|
| 133 |
+
raise TypeError("Invalid input text. Please provide a string, or a list of strings")
|
| 134 |
+
|
| 135 |
+
image_inputs = video_inputs = {}
|
| 136 |
+
|
| 137 |
+
if images is not None:
|
| 138 |
+
image_inputs = self.image_processor(images, **output_kwargs["images_kwargs"])
|
| 139 |
+
|
| 140 |
+
batch_num_images = iter(image_inputs["batch_num_images"])
|
| 141 |
+
image_sizes = iter(image_inputs["image_sizes"])
|
| 142 |
+
height, width = get_image_size(
|
| 143 |
+
to_numpy_array(image_inputs["pixel_values"][0][0]),
|
| 144 |
+
channel_dim=output_kwargs["images_kwargs"].get("data_format"),
|
| 145 |
+
)
|
| 146 |
+
text, num_image_tokens = self._expand_image_tokens(
|
| 147 |
+
text, image_sizes, height, width, self.image_token, batch_num_images
|
| 148 |
+
)
|
| 149 |
+
|
| 150 |
+
if videos is not None:
|
| 151 |
+
video_inputs = self.video_processor(videos, **output_kwargs["videos_kwargs"])
|
| 152 |
+
|
| 153 |
+
one_video = video_inputs.get("pixel_values_videos")[0]
|
| 154 |
+
if isinstance(video_inputs.get("pixel_values_videos")[0], (list, tuple)):
|
| 155 |
+
one_video = np.array(one_video)
|
| 156 |
+
else:
|
| 157 |
+
one_video = to_numpy_array(one_video)
|
| 158 |
+
num_frames = one_video.shape[0] # frame dim is always after batch dim
|
| 159 |
+
# ── ONLY deviation from upstream __call__: the video token count.
|
| 160 |
+
# Upstream derives it inline as `get_image_size(one_video[0])` (unused
|
| 161 |
+
# for count) + `num_frames * ceil(sqrt(num_image_tokens)/2)**2 + 1`.
|
| 162 |
+
# We drop the get_image_size line ON PURPOSE — the SoT helper hardcodes
|
| 163 |
+
# 144 tokens/readout-unit (= ceil(24/2)**2 for Qwen3.5 384/16 grid) and
|
| 164 |
+
# folds in the grid_t=num_frames//2 tubelet grouping when merge is on.
|
| 165 |
+
# Do NOT "restore" get_image_size when re-syncing against the default.
|
| 166 |
+
if video_mrope_active(self):
|
| 167 |
+
# FAIL-FAST: the faithful eval path builds ONE timestamp stream from a single video's
|
| 168 |
+
# (fps, sampled_indices) and replaces ONE <video> placeholder. A multi-video sample
|
| 169 |
+
# or batched (B>1) call would mis-apply the first video's timestamps to the rest.
|
| 170 |
+
# lmms-eval llava_hf runs generate at bs=1 single-video, so this never fires; it
|
| 171 |
+
# guards a future multi-video / batched entry point from silent mis-scoring.
|
| 172 |
+
n_video_ph = sum(sample.count(self.video_token) for sample in text)
|
| 173 |
+
if n_video_ph != 1:
|
| 174 |
+
raise RuntimeError(
|
| 175 |
+
f"LlavaQwen3_5ViTProcessor: faithful video_mrope eval supports exactly ONE "
|
| 176 |
+
f"<video> per call (single video, batch_size=1); found {n_video_ph}. "
|
| 177 |
+
f"Multi-video / batched faithful eval is unsupported."
|
| 178 |
+
)
|
| 179 |
+
if video_fps is None:
|
| 180 |
+
# FAIL-FAST (not a silent blob fallback): video_mrope=1 means the ckpt was trained
|
| 181 |
+
# on the faithful per-frame timestamp stream, so evaluating it on the contiguous
|
| 182 |
+
# blob is a train/eval mismatch that produces silently-wrong numbers. Only the
|
| 183 |
+
# lmms-eval simple (--force_simple) path threads fps today; the chat / VLMEvalKit
|
| 184 |
+
# video paths do not. Current 6-bench video tasks all use --force_simple, so this
|
| 185 |
+
# never fires; it guards a future entry-point swap from silently degrading.
|
| 186 |
+
# (Image-only eval never reaches here — `videos is None` skips this whole block.)
|
| 187 |
+
raise RuntimeError(
|
| 188 |
+
"LlavaQwen3_5ViTProcessor: video_mrope=1 but no video_fps/video_sample_indices "
|
| 189 |
+
"were provided — cannot build the faithful per-frame timestamp stream. Route "
|
| 190 |
+
"video eval through the lmms-eval simple (--force_simple) path, which threads fps."
|
| 191 |
+
)
|
| 192 |
+
# Faithful video-mRoPE (video_mrope=1): build the per-frame timestamp + vision-marker
|
| 193 |
+
# stream via the SAME shared SoT the training template _encode uses, so eval and
|
| 194 |
+
# train produce byte-identical video token ids (zero drift). Needs fps + sampled
|
| 195 |
+
# frame indices (threaded from the eval video loader). One video per eval request.
|
| 196 |
+
faithful_str = build_faithful_video_token_string(
|
| 197 |
+
num_frames, video_fps, video_sample_indices,
|
| 198 |
+
video_is_temporal_merge(self), video_token=self.video_token,
|
| 199 |
+
)
|
| 200 |
+
text = [sample.replace(self.video_token, faithful_str, 1) for sample in text]
|
| 201 |
+
else:
|
| 202 |
+
# Default / non-mrope: byte-identical contiguous blob.
|
| 203 |
+
num_video_tokens = qwen3_5_video_token_count(
|
| 204 |
+
num_frames,
|
| 205 |
+
temporal_merge=video_is_temporal_merge(self),
|
| 206 |
+
temporal_compression=video_is_temporal_compression(self),
|
| 207 |
+
pair_fusion=video_is_pair_fusion(self),
|
| 208 |
+
vtc_compression=video_is_vtc_compression(self),
|
| 209 |
+
spatial_catp=video_is_spatial_catp(self),
|
| 210 |
+
vtc_compression_factor=vtc_compression_factor(self),
|
| 211 |
+
vtc_temporal_patch_size=vtc_temporal_patch_size(self),
|
| 212 |
+
)
|
| 213 |
+
text = [sample.replace(self.video_token, self.video_token * num_video_tokens) for sample in text]
|
| 214 |
+
|
| 215 |
+
return_tensors = output_kwargs["text_kwargs"].pop("return_tensors", None)
|
| 216 |
+
return_mm_token_type_ids = output_kwargs["text_kwargs"].pop("return_mm_token_type_ids", None)
|
| 217 |
+
text_inputs = self.tokenizer(text, **output_kwargs["text_kwargs"])
|
| 218 |
+
self._check_special_mm_tokens(text, text_inputs, modalities=["image"])
|
| 219 |
+
|
| 220 |
+
if return_mm_token_type_ids:
|
| 221 |
+
# Byte-identical to LlavaOnevisionProcessor.__call__ (transformers 5.5.4):
|
| 222 |
+
# delegate to ProcessorMixin.create_mm_token_type_ids, which marks image=1 / video=2
|
| 223 |
+
# / audio=3 over self.image_ids/video_ids/audio_ids (set by the base __init__, so
|
| 224 |
+
# present on the fresh-build AND the __class__-retype reload path). An earlier inline
|
| 225 |
+
# copy here came from an OLDER transformers that lacked this method and marked
|
| 226 |
+
# image-only (mis-typing video tokens); 5.5.4 HAS the method (processing_utils.py) and
|
| 227 |
+
# the default path calls it, so tracking the default keeps the <video> token COUNT the only intentional
|
| 228 |
+
# deviation in this __call__. (This path is dormant: return_mm_token_type_ids defaults
|
| 229 |
+
# False for LlavaOnevision.)
|
| 230 |
+
text_inputs["mm_token_type_ids"] = self.create_mm_token_type_ids(text_inputs["input_ids"])
|
| 231 |
+
return BatchFeature(data={**text_inputs, **image_inputs, **video_inputs}, tensor_type=return_tensors)
|
| 232 |
+
|
| 233 |
+
def _get_number_of_features(self, orig_height: int, orig_width: int, height: int, width: int) -> int:
|
| 234 |
+
"""image_mrope=1: drop the per-row image_newline from the count (the model calls
|
| 235 |
+
pack_image_features with image_newline=None), so the image token count is
|
| 236 |
+
base + unpadded (clean H x W grid). Default count = base + unpadded + newline; here we
|
| 237 |
+
subtract exactly `newline_features`. flag 0 (default / video_mrope) → byte-identical to parent.
|
| 238 |
+
Used by BOTH the swift training _encode and eval __call__/_expand_image_tokens, so the
|
| 239 |
+
model's pack feature count and the input_ids placeholder count stay matched."""
|
| 240 |
+
n = super()._get_number_of_features(orig_height, orig_width, height, width)
|
| 241 |
+
if video_is_image_mrope(self):
|
| 242 |
+
# Guard the one config that would desync the position-id block sum from the emitted
|
| 243 |
+
# token count: vision_feature_select_strategy="default" makes _expand_image_tokens
|
| 244 |
+
# subtract 1 (a CLS-token drop), but image_mrope_grid_blocks emits the full base+
|
| 245 |
+
# unpadded grid (no CLS to drop). Qwen3.5 ViT has NO CLS token, so the strategy is
|
| 246 |
+
# always 'full'/None here; fail loud instead of silently misaligning if that ever changes.
|
| 247 |
+
if getattr(self, "vision_feature_select_strategy", None) == "default":
|
| 248 |
+
raise ValueError(
|
| 249 |
+
"image_mrope is incompatible with vision_feature_select_strategy='default' "
|
| 250 |
+
"(no CLS token in the clean H x W grid; the -1 would desync the position-id "
|
| 251 |
+
"block sum from the token count). Use 'full' / None."
|
| 252 |
+
)
|
| 253 |
+
image_grid_pinpoints = self.image_processor.image_grid_pinpoints
|
| 254 |
+
height_best, width_best = select_best_resolution([orig_height, orig_width], image_grid_pinpoints)
|
| 255 |
+
scale_height, scale_width = height_best // height, width_best // width
|
| 256 |
+
patches = int(math.sqrt(self.num_image_tokens))
|
| 257 |
+
_, newline_features = self._get_unpadded_features(
|
| 258 |
+
orig_height, orig_width, patches, patches, scale_height, scale_width
|
| 259 |
+
)
|
| 260 |
+
n -= newline_features
|
| 261 |
+
return n
|
| 262 |
+
|
| 263 |
+
def _expand_image_tokens(self, text, image_sizes, height, width, special_token, batch_num_images):
|
| 264 |
+
"""image_mrope=1 multi-image fix: LlavaOV expands each image of a MULTI-image sample as
|
| 265 |
+
`num_image_tokens + 1` (base 576 + one image_newline), bypassing `_get_number_of_features`
|
| 266 |
+
(HF processing_llava_onevision.py). But under image_mrope the model strips image_newline
|
| 267 |
+
(pack_image_features(image_newline=None)) → 576 features/image, so the default +1 over-counts
|
| 268 |
+
by one per image → masked_scatter token/feature mismatch at multi-image eval (e.g. MMMU).
|
| 269 |
+
This override drops the +1 ONLY for image_mrope multi-image; single-image still routes
|
| 270 |
+
through `_get_number_of_features` (AnyRes, newline already stripped there). flag 0 →
|
| 271 |
+
byte-identical to the parent. Verbatim from LlavaOnevisionProcessor._expand_image_tokens
|
| 272 |
+
EXCEPT the gated multi-image count line."""
|
| 273 |
+
image_mrope = video_is_image_mrope(self)
|
| 274 |
+
prompt_strings = []
|
| 275 |
+
max_num_vision_tokens = 0
|
| 276 |
+
for sample in text:
|
| 277 |
+
if special_token in sample:
|
| 278 |
+
num_images = next(batch_num_images) # should consume iterable
|
| 279 |
+
is_multi_image = num_images != 1
|
| 280 |
+
else:
|
| 281 |
+
is_multi_image = False
|
| 282 |
+
while special_token in sample:
|
| 283 |
+
original_size = next(image_sizes) # should consume iterable
|
| 284 |
+
if is_multi_image:
|
| 285 |
+
# ── ONLY deviation: drop the +1 newline under image_mrope (base-only 576).
|
| 286 |
+
num_image_tokens = self.num_image_tokens if image_mrope else self.num_image_tokens + 1
|
| 287 |
+
else:
|
| 288 |
+
if not isinstance(original_size, (list, tuple)):
|
| 289 |
+
# cast to list to avoid numerical precision errors when calculating unpadding
|
| 290 |
+
original_size = original_size.tolist()
|
| 291 |
+
orig_height, orig_width = original_size
|
| 292 |
+
num_image_tokens = self._get_number_of_features(orig_height, orig_width, height, width)
|
| 293 |
+
max_num_vision_tokens = max(max_num_vision_tokens, num_image_tokens)
|
| 294 |
+
if self.vision_feature_select_strategy == "default":
|
| 295 |
+
num_image_tokens -= 1
|
| 296 |
+
sample = sample.replace(special_token, "<placeholder>" * num_image_tokens, 1)
|
| 297 |
+
prompt_strings.append(sample)
|
| 298 |
+
text = [sample.replace("<placeholder>", special_token) for sample in prompt_strings]
|
| 299 |
+
return text, max_num_vision_tokens
|
| 300 |
+
|
| 301 |
+
|
| 302 |
+
__all__ = ["LlavaQwen3_5ViTProcessor"]
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/processor_config.json
ADDED
|
@@ -0,0 +1,209 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"image_processor": {
|
| 3 |
+
"do_convert_rgb": true,
|
| 4 |
+
"do_normalize": true,
|
| 5 |
+
"do_pad": true,
|
| 6 |
+
"do_rescale": true,
|
| 7 |
+
"do_resize": true,
|
| 8 |
+
"image_grid_pinpoints": [
|
| 9 |
+
[
|
| 10 |
+
384,
|
| 11 |
+
384
|
| 12 |
+
],
|
| 13 |
+
[
|
| 14 |
+
384,
|
| 15 |
+
768
|
| 16 |
+
],
|
| 17 |
+
[
|
| 18 |
+
384,
|
| 19 |
+
1152
|
| 20 |
+
],
|
| 21 |
+
[
|
| 22 |
+
384,
|
| 23 |
+
1536
|
| 24 |
+
],
|
| 25 |
+
[
|
| 26 |
+
384,
|
| 27 |
+
1920
|
| 28 |
+
],
|
| 29 |
+
[
|
| 30 |
+
384,
|
| 31 |
+
2304
|
| 32 |
+
],
|
| 33 |
+
[
|
| 34 |
+
768,
|
| 35 |
+
384
|
| 36 |
+
],
|
| 37 |
+
[
|
| 38 |
+
768,
|
| 39 |
+
768
|
| 40 |
+
],
|
| 41 |
+
[
|
| 42 |
+
768,
|
| 43 |
+
1152
|
| 44 |
+
],
|
| 45 |
+
[
|
| 46 |
+
768,
|
| 47 |
+
1536
|
| 48 |
+
],
|
| 49 |
+
[
|
| 50 |
+
768,
|
| 51 |
+
1920
|
| 52 |
+
],
|
| 53 |
+
[
|
| 54 |
+
768,
|
| 55 |
+
2304
|
| 56 |
+
],
|
| 57 |
+
[
|
| 58 |
+
1152,
|
| 59 |
+
384
|
| 60 |
+
],
|
| 61 |
+
[
|
| 62 |
+
1152,
|
| 63 |
+
768
|
| 64 |
+
],
|
| 65 |
+
[
|
| 66 |
+
1152,
|
| 67 |
+
1152
|
| 68 |
+
],
|
| 69 |
+
[
|
| 70 |
+
1152,
|
| 71 |
+
1536
|
| 72 |
+
],
|
| 73 |
+
[
|
| 74 |
+
1152,
|
| 75 |
+
1920
|
| 76 |
+
],
|
| 77 |
+
[
|
| 78 |
+
1152,
|
| 79 |
+
2304
|
| 80 |
+
],
|
| 81 |
+
[
|
| 82 |
+
1536,
|
| 83 |
+
384
|
| 84 |
+
],
|
| 85 |
+
[
|
| 86 |
+
1536,
|
| 87 |
+
768
|
| 88 |
+
],
|
| 89 |
+
[
|
| 90 |
+
1536,
|
| 91 |
+
1152
|
| 92 |
+
],
|
| 93 |
+
[
|
| 94 |
+
1536,
|
| 95 |
+
1536
|
| 96 |
+
],
|
| 97 |
+
[
|
| 98 |
+
1536,
|
| 99 |
+
1920
|
| 100 |
+
],
|
| 101 |
+
[
|
| 102 |
+
1536,
|
| 103 |
+
2304
|
| 104 |
+
],
|
| 105 |
+
[
|
| 106 |
+
1920,
|
| 107 |
+
384
|
| 108 |
+
],
|
| 109 |
+
[
|
| 110 |
+
1920,
|
| 111 |
+
768
|
| 112 |
+
],
|
| 113 |
+
[
|
| 114 |
+
1920,
|
| 115 |
+
1152
|
| 116 |
+
],
|
| 117 |
+
[
|
| 118 |
+
1920,
|
| 119 |
+
1536
|
| 120 |
+
],
|
| 121 |
+
[
|
| 122 |
+
1920,
|
| 123 |
+
1920
|
| 124 |
+
],
|
| 125 |
+
[
|
| 126 |
+
1920,
|
| 127 |
+
2304
|
| 128 |
+
],
|
| 129 |
+
[
|
| 130 |
+
2304,
|
| 131 |
+
384
|
| 132 |
+
],
|
| 133 |
+
[
|
| 134 |
+
2304,
|
| 135 |
+
768
|
| 136 |
+
],
|
| 137 |
+
[
|
| 138 |
+
2304,
|
| 139 |
+
1152
|
| 140 |
+
],
|
| 141 |
+
[
|
| 142 |
+
2304,
|
| 143 |
+
1536
|
| 144 |
+
],
|
| 145 |
+
[
|
| 146 |
+
2304,
|
| 147 |
+
1920
|
| 148 |
+
],
|
| 149 |
+
[
|
| 150 |
+
2304,
|
| 151 |
+
2304
|
| 152 |
+
]
|
| 153 |
+
],
|
| 154 |
+
"image_mean": [
|
| 155 |
+
0.5,
|
| 156 |
+
0.5,
|
| 157 |
+
0.5
|
| 158 |
+
],
|
| 159 |
+
"image_processor_type": "LlavaOnevisionImageProcessor",
|
| 160 |
+
"image_std": [
|
| 161 |
+
0.5,
|
| 162 |
+
0.5,
|
| 163 |
+
0.5
|
| 164 |
+
],
|
| 165 |
+
"resample": 3,
|
| 166 |
+
"rescale_factor": 0.00392156862745098,
|
| 167 |
+
"size": {
|
| 168 |
+
"height": 384,
|
| 169 |
+
"width": 384
|
| 170 |
+
}
|
| 171 |
+
},
|
| 172 |
+
"image_token": "<image>",
|
| 173 |
+
"num_image_tokens": 576,
|
| 174 |
+
"processor_class": "LlavaQwen3_5ViTProcessor",
|
| 175 |
+
"video_mode": "temporal_merge",
|
| 176 |
+
"video_processor": {
|
| 177 |
+
"do_convert_rgb": true,
|
| 178 |
+
"do_normalize": true,
|
| 179 |
+
"do_rescale": true,
|
| 180 |
+
"do_resize": true,
|
| 181 |
+
"do_sample_frames": false,
|
| 182 |
+
"image_mean": [
|
| 183 |
+
0.5,
|
| 184 |
+
0.5,
|
| 185 |
+
0.5
|
| 186 |
+
],
|
| 187 |
+
"image_std": [
|
| 188 |
+
0.5,
|
| 189 |
+
0.5,
|
| 190 |
+
0.5
|
| 191 |
+
],
|
| 192 |
+
"resample": 3,
|
| 193 |
+
"rescale_factor": 0.00392156862745098,
|
| 194 |
+
"return_metadata": false,
|
| 195 |
+
"size": {
|
| 196 |
+
"height": 384,
|
| 197 |
+
"width": 384
|
| 198 |
+
},
|
| 199 |
+
"video_processor_type": "LlavaOnevisionVideoProcessor"
|
| 200 |
+
},
|
| 201 |
+
"video_token": "<video>",
|
| 202 |
+
"vision_aspect_ratio": "anyres_max_9",
|
| 203 |
+
"vision_feature_select_strategy": null,
|
| 204 |
+
"vtc_compression_factor": 2,
|
| 205 |
+
"vtc_temporal_patch_size": 1,
|
| 206 |
+
"auto_map": {
|
| 207 |
+
"AutoProcessor": "processing_qwen3_5vit_qwen3.LlavaQwen3_5ViTProcessor"
|
| 208 |
+
}
|
| 209 |
+
}
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/tokenizer.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9024318c850eaadf26be79389d21b07a7afd8f1af749b89f9109b06c0d12173c
|
| 3 |
+
size 11423018
|
ckpts/S2/8b/qwen3_5_2b/baseline_tmerge_64f_s2_lora/v0-20261021-011325/checkpoint-1533-merged/tokenizer_config.json
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": null,
|
| 5 |
+
"clean_up_tokenization_spaces": false,
|
| 6 |
+
"eos_token": "<|im_end|>",
|
| 7 |
+
"errors": "replace",
|
| 8 |
+
"extra_special_tokens": [
|
| 9 |
+
"<image>",
|
| 10 |
+
"<video>"
|
| 11 |
+
],
|
| 12 |
+
"is_local": true,
|
| 13 |
+
"model_max_length": 131072,
|
| 14 |
+
"pad_token": "<|endoftext|>",
|
| 15 |
+
"processor_class": "LlavaQwen3_5ViTProcessor",
|
| 16 |
+
"split_special_tokens": false,
|
| 17 |
+
"tokenizer_class": "Qwen2Tokenizer",
|
| 18 |
+
"unk_token": null
|
| 19 |
+
}
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/args.json
ADDED
|
@@ -0,0 +1,393 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"output_dir": "/home/wangjunjie09/VisionEncoder/V3_exps/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726",
|
| 3 |
+
"per_device_train_batch_size": 1,
|
| 4 |
+
"num_train_epochs": 1.0,
|
| 5 |
+
"max_steps": -1,
|
| 6 |
+
"learning_rate": 0.0001,
|
| 7 |
+
"lr_scheduler_type": "cosine",
|
| 8 |
+
"lr_scheduler_kwargs": null,
|
| 9 |
+
"warmup_steps": 0,
|
| 10 |
+
"optim": "adamw_torch_fused",
|
| 11 |
+
"optim_args": null,
|
| 12 |
+
"weight_decay": 0.1,
|
| 13 |
+
"adam_beta1": 0.9,
|
| 14 |
+
"adam_beta2": 0.95,
|
| 15 |
+
"adam_epsilon": 1e-08,
|
| 16 |
+
"optim_target_modules": null,
|
| 17 |
+
"gradient_accumulation_steps": 8,
|
| 18 |
+
"average_tokens_across_devices": true,
|
| 19 |
+
"max_grad_norm": 1.0,
|
| 20 |
+
"label_smoothing_factor": 0.0,
|
| 21 |
+
"bf16": true,
|
| 22 |
+
"fp16": false,
|
| 23 |
+
"bf16_full_eval": false,
|
| 24 |
+
"fp16_full_eval": false,
|
| 25 |
+
"tf32": null,
|
| 26 |
+
"gradient_checkpointing": true,
|
| 27 |
+
"gradient_checkpointing_kwargs": "{\"use_reentrant\": false}",
|
| 28 |
+
"torch_compile": false,
|
| 29 |
+
"torch_compile_backend": null,
|
| 30 |
+
"torch_compile_mode": null,
|
| 31 |
+
"use_liger_kernel": false,
|
| 32 |
+
"liger_kernel_config": null,
|
| 33 |
+
"use_cache": false,
|
| 34 |
+
"neftune_noise_alpha": null,
|
| 35 |
+
"torch_empty_cache_steps": null,
|
| 36 |
+
"auto_find_batch_size": false,
|
| 37 |
+
"logging_strategy": "steps",
|
| 38 |
+
"logging_steps": 5,
|
| 39 |
+
"logging_first_step": true,
|
| 40 |
+
"log_on_each_node": true,
|
| 41 |
+
"logging_nan_inf_filter": true,
|
| 42 |
+
"include_num_input_tokens_seen": false,
|
| 43 |
+
"log_level": "passive",
|
| 44 |
+
"log_level_replica": "warning",
|
| 45 |
+
"disable_tqdm": null,
|
| 46 |
+
"report_to": [
|
| 47 |
+
"none"
|
| 48 |
+
],
|
| 49 |
+
"run_name": "/home/wangjunjie09/exps/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726",
|
| 50 |
+
"project": "huggingface",
|
| 51 |
+
"trackio_space_id": "trackio",
|
| 52 |
+
"eval_strategy": "no",
|
| 53 |
+
"eval_steps": 100.0,
|
| 54 |
+
"eval_delay": 0,
|
| 55 |
+
"per_device_eval_batch_size": 1,
|
| 56 |
+
"prediction_loss_only": false,
|
| 57 |
+
"eval_on_start": false,
|
| 58 |
+
"eval_do_concat_batches": true,
|
| 59 |
+
"eval_use_gather_object": false,
|
| 60 |
+
"eval_accumulation_steps": null,
|
| 61 |
+
"include_for_metrics": [],
|
| 62 |
+
"batch_eval_metrics": false,
|
| 63 |
+
"save_only_model": false,
|
| 64 |
+
"save_strategy": "steps",
|
| 65 |
+
"save_steps": 100.0,
|
| 66 |
+
"save_on_each_node": false,
|
| 67 |
+
"save_total_limit": 1,
|
| 68 |
+
"enable_jit_checkpoint": false,
|
| 69 |
+
"push_to_hub": false,
|
| 70 |
+
"hub_token": null,
|
| 71 |
+
"hub_private_repo": null,
|
| 72 |
+
"hub_model_id": null,
|
| 73 |
+
"hub_strategy": "every_save",
|
| 74 |
+
"hub_always_push": false,
|
| 75 |
+
"hub_revision": null,
|
| 76 |
+
"load_best_model_at_end": false,
|
| 77 |
+
"metric_for_best_model": "loss",
|
| 78 |
+
"greater_is_better": false,
|
| 79 |
+
"ignore_data_skip": false,
|
| 80 |
+
"restore_callback_states_from_checkpoint": false,
|
| 81 |
+
"full_determinism": false,
|
| 82 |
+
"seed": 42,
|
| 83 |
+
"data_seed": 42,
|
| 84 |
+
"use_cpu": false,
|
| 85 |
+
"accelerator_config": "{\"dispatch_batches\": false}",
|
| 86 |
+
"parallelism_config": null,
|
| 87 |
+
"dataloader_drop_last": false,
|
| 88 |
+
"dataloader_num_workers": 8,
|
| 89 |
+
"dataloader_pin_memory": true,
|
| 90 |
+
"dataloader_persistent_workers": true,
|
| 91 |
+
"dataloader_prefetch_factor": 4,
|
| 92 |
+
"remove_unused_columns": true,
|
| 93 |
+
"label_names": null,
|
| 94 |
+
"train_sampling_strategy": "random",
|
| 95 |
+
"length_column_name": "length",
|
| 96 |
+
"ddp_find_unused_parameters": null,
|
| 97 |
+
"ddp_bucket_cap_mb": null,
|
| 98 |
+
"ddp_broadcast_buffers": null,
|
| 99 |
+
"ddp_backend": null,
|
| 100 |
+
"ddp_timeout": 7200,
|
| 101 |
+
"fsdp": [],
|
| 102 |
+
"fsdp_config": null,
|
| 103 |
+
"deepspeed": {
|
| 104 |
+
"fp16": {
|
| 105 |
+
"enabled": "auto",
|
| 106 |
+
"loss_scale": 0,
|
| 107 |
+
"loss_scale_window": 1000,
|
| 108 |
+
"initial_scale_power": 16,
|
| 109 |
+
"hysteresis": 2,
|
| 110 |
+
"min_loss_scale": 1
|
| 111 |
+
},
|
| 112 |
+
"bf16": {
|
| 113 |
+
"enabled": "auto"
|
| 114 |
+
},
|
| 115 |
+
"zero_optimization": {
|
| 116 |
+
"stage": 2,
|
| 117 |
+
"offload_optimizer": {
|
| 118 |
+
"device": "none",
|
| 119 |
+
"pin_memory": true
|
| 120 |
+
},
|
| 121 |
+
"allgather_partitions": true,
|
| 122 |
+
"allgather_bucket_size": 200000000.0,
|
| 123 |
+
"overlap_comm": false,
|
| 124 |
+
"reduce_scatter": true,
|
| 125 |
+
"reduce_bucket_size": 200000000.0,
|
| 126 |
+
"contiguous_gradients": true
|
| 127 |
+
},
|
| 128 |
+
"gradient_accumulation_steps": "auto",
|
| 129 |
+
"gradient_clipping": "auto",
|
| 130 |
+
"steps_per_print": 2000,
|
| 131 |
+
"train_batch_size": "auto",
|
| 132 |
+
"train_micro_batch_size_per_gpu": "auto",
|
| 133 |
+
"wall_clock_breakdown": false
|
| 134 |
+
},
|
| 135 |
+
"debug": null,
|
| 136 |
+
"skip_memory_metrics": true,
|
| 137 |
+
"do_train": false,
|
| 138 |
+
"do_eval": false,
|
| 139 |
+
"do_predict": false,
|
| 140 |
+
"resume_from_checkpoint": "/home/wangjunjie09/exps/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v1-20261025-030229/checkpoint-1000",
|
| 141 |
+
"warmup_ratio": 0.05,
|
| 142 |
+
"logging_dir": "/home/wangjunjie09/exps/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/runs",
|
| 143 |
+
"local_rank": 0,
|
| 144 |
+
"sortish_sampler": false,
|
| 145 |
+
"predict_with_generate": false,
|
| 146 |
+
"generation_max_length": null,
|
| 147 |
+
"generation_num_beams": null,
|
| 148 |
+
"generation_config": null,
|
| 149 |
+
"tuner_backend": "peft",
|
| 150 |
+
"llm_checkpointing": 12,
|
| 151 |
+
"vit_gradient_checkpointing": true,
|
| 152 |
+
"router_aux_loss_coef": 0.0,
|
| 153 |
+
"enable_dft_loss": false,
|
| 154 |
+
"enable_channel_loss": false,
|
| 155 |
+
"safe_serialization": true,
|
| 156 |
+
"max_shard_size": "5GB",
|
| 157 |
+
"check_model": true,
|
| 158 |
+
"acc_strategy": "token",
|
| 159 |
+
"train_dataloader_shuffle": true,
|
| 160 |
+
"group_by_length": false,
|
| 161 |
+
"group_by_packed_video_cost": true,
|
| 162 |
+
"group_by_packed_compute_cost": false,
|
| 163 |
+
"group_by_packed_modality": false,
|
| 164 |
+
"max_epochs": null,
|
| 165 |
+
"aligner_lr": 1e-05,
|
| 166 |
+
"vit_lr": 1e-06,
|
| 167 |
+
"use_logits_to_keep": true,
|
| 168 |
+
"ds3_gather_for_generation": true,
|
| 169 |
+
"resume_only_model": false,
|
| 170 |
+
"optimizer": null,
|
| 171 |
+
"loss_type": null,
|
| 172 |
+
"eval_metric": null,
|
| 173 |
+
"callbacks": [],
|
| 174 |
+
"early_stop_interval": null,
|
| 175 |
+
"eval_use_evalscope": false,
|
| 176 |
+
"eval_dataset": [],
|
| 177 |
+
"eval_dataset_args": null,
|
| 178 |
+
"eval_limit": null,
|
| 179 |
+
"eval_generation_config": null,
|
| 180 |
+
"extra_eval_args": null,
|
| 181 |
+
"tuner_type": "lora",
|
| 182 |
+
"use_galore": false,
|
| 183 |
+
"galore_target_modules": null,
|
| 184 |
+
"galore_rank": 128,
|
| 185 |
+
"galore_update_proj_gap": 50,
|
| 186 |
+
"galore_scale": 1.0,
|
| 187 |
+
"galore_proj_type": "std",
|
| 188 |
+
"galore_optim_per_parameter": false,
|
| 189 |
+
"galore_with_embedding": false,
|
| 190 |
+
"galore_quantization": false,
|
| 191 |
+
"galore_proj_quant": false,
|
| 192 |
+
"galore_proj_bits": 4,
|
| 193 |
+
"galore_proj_group_size": 256,
|
| 194 |
+
"galore_cos_threshold": 0.4,
|
| 195 |
+
"galore_gamma_proj": 2,
|
| 196 |
+
"galore_queue_size": 5,
|
| 197 |
+
"lisa_activated_layers": 0,
|
| 198 |
+
"lisa_step_interval": 20,
|
| 199 |
+
"use_flash_ckpt": false,
|
| 200 |
+
"use_ray": false,
|
| 201 |
+
"ray_exp_name": null,
|
| 202 |
+
"device_groups": null,
|
| 203 |
+
"model": "/home/wangjunjie09/VisionEncoder/V3_exps/S1/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v0-20261023-060432/checkpoint-2181",
|
| 204 |
+
"model_type": "llava_qwen3_5vit_qwen3",
|
| 205 |
+
"model_revision": null,
|
| 206 |
+
"task_type": "causal_lm",
|
| 207 |
+
"torch_dtype": "bfloat16",
|
| 208 |
+
"attn_impl": "flash_attn",
|
| 209 |
+
"experts_impl": null,
|
| 210 |
+
"new_special_tokens": [],
|
| 211 |
+
"num_labels": null,
|
| 212 |
+
"problem_type": null,
|
| 213 |
+
"rope_scaling": null,
|
| 214 |
+
"device_map": null,
|
| 215 |
+
"max_memory": {},
|
| 216 |
+
"max_model_len": null,
|
| 217 |
+
"local_repo_path": null,
|
| 218 |
+
"init_strategy": null,
|
| 219 |
+
"template": "llava_qwen3_5vit_qwen3_nonthinking",
|
| 220 |
+
"system": null,
|
| 221 |
+
"max_length": 16384,
|
| 222 |
+
"truncation_strategy": "delete",
|
| 223 |
+
"max_pixels": null,
|
| 224 |
+
"agent_template": null,
|
| 225 |
+
"norm_bbox": null,
|
| 226 |
+
"use_chat_template": true,
|
| 227 |
+
"padding_side": "right",
|
| 228 |
+
"padding_free": true,
|
| 229 |
+
"loss_scale": "default+ignore_empty_think",
|
| 230 |
+
"sequence_parallel_size": 1,
|
| 231 |
+
"template_backend": "swift",
|
| 232 |
+
"response_prefix": null,
|
| 233 |
+
"enable_thinking": false,
|
| 234 |
+
"add_non_thinking_prefix": true,
|
| 235 |
+
"dataset": [],
|
| 236 |
+
"val_dataset": [],
|
| 237 |
+
"cached_dataset": [
|
| 238 |
+
"/share/m2v_intern/wangjunjie09/datasets/prepared/s2/qwen3_8b/image/train",
|
| 239 |
+
"/share/m2v_intern/wangjunjie09/datasets/prepared/s2/qwen3_8b/video/train"
|
| 240 |
+
],
|
| 241 |
+
"cached_val_dataset": [],
|
| 242 |
+
"split_dataset_ratio": 0.0,
|
| 243 |
+
"dataset_num_proc": 16,
|
| 244 |
+
"load_from_cache_file": false,
|
| 245 |
+
"dataset_shuffle": true,
|
| 246 |
+
"val_dataset_shuffle": false,
|
| 247 |
+
"streaming": false,
|
| 248 |
+
"interleave_prob": null,
|
| 249 |
+
"stopping_strategy": "first_exhausted",
|
| 250 |
+
"shuffle_buffer_size": 1000,
|
| 251 |
+
"download_mode": "reuse_dataset_if_exists",
|
| 252 |
+
"columns": {},
|
| 253 |
+
"strict": false,
|
| 254 |
+
"model_name": null,
|
| 255 |
+
"model_author": null,
|
| 256 |
+
"custom_dataset_info": [],
|
| 257 |
+
"quant_method": null,
|
| 258 |
+
"quant_bits": null,
|
| 259 |
+
"hqq_axis": null,
|
| 260 |
+
"bnb_4bit_compute_dtype": "bfloat16",
|
| 261 |
+
"bnb_4bit_quant_type": "nf4",
|
| 262 |
+
"bnb_4bit_use_double_quant": true,
|
| 263 |
+
"bnb_4bit_quant_storage": null,
|
| 264 |
+
"max_new_tokens": 64,
|
| 265 |
+
"temperature": 0.0,
|
| 266 |
+
"top_k": null,
|
| 267 |
+
"top_p": null,
|
| 268 |
+
"repetition_penalty": null,
|
| 269 |
+
"num_beams": 1,
|
| 270 |
+
"stream": false,
|
| 271 |
+
"stop_words": [],
|
| 272 |
+
"logprobs": false,
|
| 273 |
+
"top_logprobs": null,
|
| 274 |
+
"structured_outputs_regex": null,
|
| 275 |
+
"train_type": null,
|
| 276 |
+
"adapters": [],
|
| 277 |
+
"external_plugins": [
|
| 278 |
+
"video_mllm/training/model_plugin_qwen3_5vit.py",
|
| 279 |
+
"video_mllm/data/dataset_plugin.py",
|
| 280 |
+
"/home/wangjunjie09/VisionEncoder/video_mllm/training/log_spacing_plugin.py"
|
| 281 |
+
],
|
| 282 |
+
"custom_register_path": [
|
| 283 |
+
"/home/wangjunjie09/VisionEncoder/video_mllm/training/log_spacing_plugin.py"
|
| 284 |
+
],
|
| 285 |
+
"model_kwargs": {
|
| 286 |
+
"vision_source_id": "qwen3_5_2b",
|
| 287 |
+
"qwen3_5_vit_video_mode": "vtc_compression",
|
| 288 |
+
"num_frames": 64,
|
| 289 |
+
"eval_frames": 64,
|
| 290 |
+
"s2_use_video_frame_cache": true,
|
| 291 |
+
"s2_video_frame_cache": "/share/m2v_intern/wangjunjie09/datasets/video_cache/s2/frames/jpeg/384/64"
|
| 292 |
+
},
|
| 293 |
+
"video_cache": {},
|
| 294 |
+
"image_teacher_models": {},
|
| 295 |
+
"load_args": false,
|
| 296 |
+
"load_data_args": false,
|
| 297 |
+
"packing": true,
|
| 298 |
+
"packing_length": 16384,
|
| 299 |
+
"packing_num_proc": 1,
|
| 300 |
+
"lazy_tokenize": false,
|
| 301 |
+
"use_hf": true,
|
| 302 |
+
"ignore_args_error": false,
|
| 303 |
+
"use_swift_lora": false,
|
| 304 |
+
"freeze_parameters": [],
|
| 305 |
+
"freeze_parameters_regex": null,
|
| 306 |
+
"freeze_parameters_ratio": 0.0,
|
| 307 |
+
"trainable_parameters": [],
|
| 308 |
+
"trainable_parameters_regex": null,
|
| 309 |
+
"freeze_llm": false,
|
| 310 |
+
"freeze_vit": false,
|
| 311 |
+
"freeze_aligner": false,
|
| 312 |
+
"target_modules": "model\\.language_model\\.layers\\.[0-9]+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))",
|
| 313 |
+
"target_regex": "model\\.language_model\\.layers\\.[0-9]+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))",
|
| 314 |
+
"target_parameters": null,
|
| 315 |
+
"modules_to_save": [
|
| 316 |
+
"vision_tower",
|
| 317 |
+
"multi_modal_projector"
|
| 318 |
+
],
|
| 319 |
+
"lora_rank": 128,
|
| 320 |
+
"lora_alpha": 256,
|
| 321 |
+
"lora_dropout": 0.0,
|
| 322 |
+
"lora_bias": "none",
|
| 323 |
+
"lora_dtype": null,
|
| 324 |
+
"lorap_lr_ratio": null,
|
| 325 |
+
"use_rslora": false,
|
| 326 |
+
"use_dora": false,
|
| 327 |
+
"lora_ga_batch_size": 2,
|
| 328 |
+
"lora_ga_iters": 2,
|
| 329 |
+
"lora_ga_max_length": 1024,
|
| 330 |
+
"lora_ga_direction": "ArB2r",
|
| 331 |
+
"lora_ga_scale": "stable",
|
| 332 |
+
"lora_ga_stable_gamma": 16,
|
| 333 |
+
"init_weights": true,
|
| 334 |
+
"fourier_n_frequency": 2000,
|
| 335 |
+
"fourier_scaling": 300.0,
|
| 336 |
+
"boft_block_size": 4,
|
| 337 |
+
"boft_block_num": 0,
|
| 338 |
+
"boft_n_butterfly_factor": 1,
|
| 339 |
+
"boft_dropout": 0.0,
|
| 340 |
+
"vera_rank": 256,
|
| 341 |
+
"vera_projection_prng_key": 0,
|
| 342 |
+
"vera_dropout": 0.0,
|
| 343 |
+
"vera_d_initial": 0.1,
|
| 344 |
+
"adapter_act": "gelu",
|
| 345 |
+
"adapter_length": 128,
|
| 346 |
+
"adalora_target_r": 8,
|
| 347 |
+
"adalora_init_r": 12,
|
| 348 |
+
"adalora_tinit": 0,
|
| 349 |
+
"adalora_tfinal": 0,
|
| 350 |
+
"adalora_deltaT": 1,
|
| 351 |
+
"adalora_beta1": 0.85,
|
| 352 |
+
"adalora_beta2": 0.85,
|
| 353 |
+
"adalora_orth_reg_weight": 0.5,
|
| 354 |
+
"llamapro_num_new_blocks": 4,
|
| 355 |
+
"llamapro_num_groups": null,
|
| 356 |
+
"reft_layer_key": null,
|
| 357 |
+
"reft_layers": null,
|
| 358 |
+
"reft_rank": 4,
|
| 359 |
+
"reft_intervention_type": "LoreftIntervention",
|
| 360 |
+
"reft_args": null,
|
| 361 |
+
"swanlab_token": null,
|
| 362 |
+
"swanlab_project": "ms-swift",
|
| 363 |
+
"swanlab_workspace": null,
|
| 364 |
+
"swanlab_exp_name": null,
|
| 365 |
+
"swanlab_notification_method": null,
|
| 366 |
+
"swanlab_webhook_url": null,
|
| 367 |
+
"swanlab_secret": null,
|
| 368 |
+
"swanlab_sender_email": null,
|
| 369 |
+
"swanlab_receiver_email": null,
|
| 370 |
+
"swanlab_smtp_server": null,
|
| 371 |
+
"swanlab_smtp_port": null,
|
| 372 |
+
"swanlab_email_language": "zh",
|
| 373 |
+
"swanlab_mode": "cloud",
|
| 374 |
+
"add_version": true,
|
| 375 |
+
"create_checkpoint_symlink": false,
|
| 376 |
+
"v12_resolution": 384,
|
| 377 |
+
"zero_hpz_partition_size": null,
|
| 378 |
+
"deepspeed_autotp_size": null,
|
| 379 |
+
"swift_version": "4.1.0.dev0",
|
| 380 |
+
"ckpt_dir": "/home/wangjunjie09/VisionEncoder/V3_exps/S1/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v0-20261023-060432/checkpoint-2181",
|
| 381 |
+
"rank": 0,
|
| 382 |
+
"global_world_size": 8,
|
| 383 |
+
"local_world_size": 8,
|
| 384 |
+
"model_suffix": "checkpoint-2181",
|
| 385 |
+
"model_info": "ModelInfo(model_type='llava_qwen3_5vit_qwen3', model_dir='/home/wangjunjie09/exps/S1/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v0-20261023-060432/checkpoint-2181', torch_dtype=torch.bfloat16, max_model_len=40960, quant_method=None, quant_bits=None, rope_scaling=None, is_moe_model=False, is_multimodal=True, config=None, task_type='causal_lm', num_labels=None)",
|
| 386 |
+
"model_meta": "ModelMeta(model_type='llava_qwen3_5vit_qwen3', model_groups=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen3-0.6B', hf_model_id='Qwen/Qwen3-0.6B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-1.7B', hf_model_id='Qwen/Qwen3-1.7B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-4B', hf_model_id='Qwen/Qwen3-4B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-8B', hf_model_id='Qwen/Qwen3-8B', model_path=None, ms_revision=None, hf_revision=None)], template=None, ignore_patterns=None, requires=None, tags=[])], loader=<class 'model_plugin_qwen3_5vit.Qwen3_5ViTLoader'>, template='llava_qwen3_5vit_qwen3', model_arch=MultiModelKeys(arch_name='llava_hf', embedding=None, module_list=None, lm_head=None, q_proj=None, k_proj=None, v_proj=None, o_proj=None, attention=None, mlp=None, down_proj=None, qkv_proj=None, qk_proj=None, qa_proj=None, qb_proj=None, kv_proj=None, kva_proj=None, kvb_proj=None, language_model=['model.language_model', 'lm_head'], aligner=['model.multi_modal_projector'], vision_tower=['model.vision_tower'], generator=[]), architectures=['LlavaQwen3_5ViTForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=[], tags=['vision', 'video'])",
|
| 387 |
+
"model_dir": "/home/wangjunjie09/VisionEncoder/V3_exps/S1/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v0-20261023-060432/checkpoint-2181",
|
| 388 |
+
"template_meta": "Qwen3MixedTemplateMeta(template_type='llava_qwen3_5vit_qwen3_nonthinking', prefix=[], prompt=['<|im_start|>user\\n{{QUERY}}<|im_end|>\\n<|im_start|>assistant\\n'], chat_sep=['<|im_end|>\\n'], suffix=['<|im_end|>\\n'], template_cls=<class 'video_mllm.training.template_qwen3_5vit.Qwen3_5ViTLlavaTemplate'>, system_prefix=['<|im_start|>system\\n{{SYSTEM}}<|im_end|>\\n'], default_system=None, auto_add_bos=False, stop_words=['<|endoftext|>'], agent_template='hermes', is_thinking=False, thinking_prefix='', non_thinking_prefix='<think>\\n\\n</think>\\n\\n', history_thinking_prefix='')",
|
| 389 |
+
"_val_dataset_exists": false,
|
| 390 |
+
"hub": "<class 'swift.hub.hub.HFHub'>",
|
| 391 |
+
"evaluation_strategy": "steps",
|
| 392 |
+
"training_args": "Seq2SeqTrainingArguments(output_dir='/home/wangjunjie09/exps/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726', per_device_train_batch_size=1, num_train_epochs=1.0, max_steps=-1, learning_rate=0.0001, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_steps=0.05, optim=<OptimizerNames.ADAMW_TORCH_FUSED: 'adamw_torch_fused'>, optim_args=None, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, optim_target_modules=None, gradient_accumulation_steps=8, average_tokens_across_devices=None, max_grad_norm=1.0, label_smoothing_factor=0.0, bf16=True, fp16=False, bf16_full_eval=False, fp16_full_eval=False, tf32=None, gradient_checkpointing=True, gradient_checkpointing_kwargs={'use_reentrant': False}, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, use_liger_kernel=False, liger_kernel_config=None, use_cache=False, neftune_noise_alpha=None, torch_empty_cache_steps=None, auto_find_batch_size=False, logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_steps=5, logging_first_step=True, log_on_each_node=True, logging_nan_inf_filter=True, include_num_input_tokens_seen=None, log_level='passive', log_level_replica='warning', disable_tqdm=False, report_to=[], run_name='/home/wangjunjie09/exps/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726', project='huggingface', trackio_space_id='trackio', eval_strategy=<IntervalStrategy.NO: 'no'>, eval_steps=100.0, eval_delay=0, per_device_eval_batch_size=1, prediction_loss_only=False, eval_on_start=False, eval_do_concat_batches=True, eval_use_gather_object=False, eval_accumulation_steps=None, include_for_metrics=[], batch_eval_metrics=False, save_only_model=False, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=100, save_on_each_node=False, save_total_limit=1, enable_jit_checkpoint=False, push_to_hub=False, hub_token=None, hub_private_repo=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_always_push=False, hub_revision=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, restore_callback_states_from_checkpoint=False, full_determinism=False, seed=42, data_seed=42, use_cpu=False, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), parallelism_config=None, dataloader_drop_last=False, dataloader_num_workers=8, dataloader_pin_memory=True, dataloader_persistent_workers=True, dataloader_prefetch_factor=4, remove_unused_columns=False, label_names=None, train_sampling_strategy='random', length_column_name='length', ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, ddp_backend=None, ddp_timeout=7200, fsdp=[], fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, deepspeed={'fp16': {'enabled': 'auto', 'loss_scale': 0, 'loss_scale_window': 1000, 'initial_scale_power': 16, 'hysteresis': 2, 'min_loss_scale': 1}, 'bf16': {'enabled': 'auto'}, 'zero_optimization': {'stage': 2, 'offload_optimizer': {'device': 'none', 'pin_memory': True}, 'allgather_partitions': True, 'allgather_bucket_size': 200000000.0, 'overlap_comm': False, 'reduce_scatter': True, 'reduce_bucket_size': 200000000.0, 'contiguous_gradients': True}, 'gradient_accumulation_steps': 'auto', 'gradient_clipping': 'auto', 'steps_per_print': 2000, 'train_batch_size': 'auto', 'train_micro_batch_size_per_gpu': 'auto', 'wall_clock_breakdown': False}, debug=[], skip_memory_metrics=True, do_train=False, do_eval=False, do_predict=False, resume_from_checkpoint='/home/wangjunjie09/exps/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v1-20261025-030229/checkpoint-1000', warmup_ratio=0.05, logging_dir='/home/wangjunjie09/exps/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/runs', local_rank=0, sortish_sampler=False, predict_with_generate=False, generation_max_length=None, generation_num_beams=None, generation_config=None, tuner_backend='peft', llm_checkpointing=12, vit_gradient_checkpointing=True, router_aux_loss_coef=0.0, enable_dft_loss=False, enable_channel_loss=False, safe_serialization=True, max_shard_size='5GB', check_model=True, acc_strategy='token', train_dataloader_shuffle=True, group_by_length=False, group_by_packed_video_cost=True, group_by_packed_compute_cost=False, group_by_packed_modality=False, max_epochs=None, aligner_lr=1e-05, vit_lr=1e-06, use_logits_to_keep=True, ds3_gather_for_generation=True, resume_only_model=False, optimizer='multimodal', loss_type=None, eval_metric=None, callbacks=[], early_stop_interval=None, eval_use_evalscope=False, eval_dataset=[], eval_dataset_args=None, eval_limit=None, eval_generation_config=None, extra_eval_args=None, tuner_type='lora', use_galore=False, galore_target_modules=None, galore_rank=128, galore_update_proj_gap=50, galore_scale=1.0, galore_proj_type='std', galore_optim_per_parameter=False, galore_with_embedding=False, galore_quantization=False, galore_proj_quant=False, galore_proj_bits=4, galore_proj_group_size=256, galore_cos_threshold=0.4, galore_gamma_proj=2, galore_queue_size=5, lisa_activated_layers=0, lisa_step_interval=20, use_flash_ckpt=False)"
|
| 393 |
+
}
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/chat_template.jinja
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{% for message in messages %}{{ '<|im_start|>' + message['role'] + '\n' }}{% if message['role'] == 'assistant' %}{{ '<think>\n\n</think>\n\n' }}{% endif %}{% if message['content'] is string %}{{ message['content'] }}{% else %}{% for content in message['content'] | selectattr('type', 'equalto', 'image') %}{{ '<image>\n' }}{% endfor %}{% for content in message['content'] | selectattr('type', 'equalto', 'video') %}{{ '<video>\n' }}{% endfor %}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{{ content['text'] }}{% endfor %}{% endif %}{{ '<|im_end|>\n' }}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n<think>\n\n</think>\n\n' }}{% endif %}
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/chat_template.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"chat_template": "{% for message in messages %}{{ '<|im_start|>' + message['role'] + '\\n' }}{% if message['role'] == 'assistant' %}{{ '<think>\\n\\n</think>\\n\\n' }}{% endif %}{% if message['content'] is string %}{{ message['content'] }}{% else %}{% for content in message['content'] | selectattr('type', 'equalto', 'image') %}{{ '<image>\\n' }}{% endfor %}{% for content in message['content'] | selectattr('type', 'equalto', 'video') %}{{ '<video>\\n' }}{% endfor %}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{{ content['text'] }}{% endfor %}{% endif %}{{ '<|im_end|>\\n' }}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\\n<think>\\n\\n</think>\\n\\n' }}{% endif %}"
|
| 3 |
+
}
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/config.json
ADDED
|
@@ -0,0 +1,289 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"LlavaQwen3_5ViTForConditionalGeneration"
|
| 4 |
+
],
|
| 5 |
+
"bos_token_id": null,
|
| 6 |
+
"dtype": "bfloat16",
|
| 7 |
+
"eos_token_id": 151645,
|
| 8 |
+
"eval_num_frames": 64,
|
| 9 |
+
"hidden_size": 4096,
|
| 10 |
+
"image_grid_pinpoints": [
|
| 11 |
+
[
|
| 12 |
+
384,
|
| 13 |
+
384
|
| 14 |
+
],
|
| 15 |
+
[
|
| 16 |
+
384,
|
| 17 |
+
768
|
| 18 |
+
],
|
| 19 |
+
[
|
| 20 |
+
384,
|
| 21 |
+
1152
|
| 22 |
+
],
|
| 23 |
+
[
|
| 24 |
+
384,
|
| 25 |
+
1536
|
| 26 |
+
],
|
| 27 |
+
[
|
| 28 |
+
384,
|
| 29 |
+
1920
|
| 30 |
+
],
|
| 31 |
+
[
|
| 32 |
+
384,
|
| 33 |
+
2304
|
| 34 |
+
],
|
| 35 |
+
[
|
| 36 |
+
768,
|
| 37 |
+
384
|
| 38 |
+
],
|
| 39 |
+
[
|
| 40 |
+
768,
|
| 41 |
+
768
|
| 42 |
+
],
|
| 43 |
+
[
|
| 44 |
+
768,
|
| 45 |
+
1152
|
| 46 |
+
],
|
| 47 |
+
[
|
| 48 |
+
768,
|
| 49 |
+
1536
|
| 50 |
+
],
|
| 51 |
+
[
|
| 52 |
+
768,
|
| 53 |
+
1920
|
| 54 |
+
],
|
| 55 |
+
[
|
| 56 |
+
768,
|
| 57 |
+
2304
|
| 58 |
+
],
|
| 59 |
+
[
|
| 60 |
+
1152,
|
| 61 |
+
384
|
| 62 |
+
],
|
| 63 |
+
[
|
| 64 |
+
1152,
|
| 65 |
+
768
|
| 66 |
+
],
|
| 67 |
+
[
|
| 68 |
+
1152,
|
| 69 |
+
1152
|
| 70 |
+
],
|
| 71 |
+
[
|
| 72 |
+
1152,
|
| 73 |
+
1536
|
| 74 |
+
],
|
| 75 |
+
[
|
| 76 |
+
1152,
|
| 77 |
+
1920
|
| 78 |
+
],
|
| 79 |
+
[
|
| 80 |
+
1152,
|
| 81 |
+
2304
|
| 82 |
+
],
|
| 83 |
+
[
|
| 84 |
+
1536,
|
| 85 |
+
384
|
| 86 |
+
],
|
| 87 |
+
[
|
| 88 |
+
1536,
|
| 89 |
+
768
|
| 90 |
+
],
|
| 91 |
+
[
|
| 92 |
+
1536,
|
| 93 |
+
1152
|
| 94 |
+
],
|
| 95 |
+
[
|
| 96 |
+
1536,
|
| 97 |
+
1536
|
| 98 |
+
],
|
| 99 |
+
[
|
| 100 |
+
1536,
|
| 101 |
+
1920
|
| 102 |
+
],
|
| 103 |
+
[
|
| 104 |
+
1536,
|
| 105 |
+
2304
|
| 106 |
+
],
|
| 107 |
+
[
|
| 108 |
+
1920,
|
| 109 |
+
384
|
| 110 |
+
],
|
| 111 |
+
[
|
| 112 |
+
1920,
|
| 113 |
+
768
|
| 114 |
+
],
|
| 115 |
+
[
|
| 116 |
+
1920,
|
| 117 |
+
1152
|
| 118 |
+
],
|
| 119 |
+
[
|
| 120 |
+
1920,
|
| 121 |
+
1536
|
| 122 |
+
],
|
| 123 |
+
[
|
| 124 |
+
1920,
|
| 125 |
+
1920
|
| 126 |
+
],
|
| 127 |
+
[
|
| 128 |
+
1920,
|
| 129 |
+
2304
|
| 130 |
+
],
|
| 131 |
+
[
|
| 132 |
+
2304,
|
| 133 |
+
384
|
| 134 |
+
],
|
| 135 |
+
[
|
| 136 |
+
2304,
|
| 137 |
+
768
|
| 138 |
+
],
|
| 139 |
+
[
|
| 140 |
+
2304,
|
| 141 |
+
1152
|
| 142 |
+
],
|
| 143 |
+
[
|
| 144 |
+
2304,
|
| 145 |
+
1536
|
| 146 |
+
],
|
| 147 |
+
[
|
| 148 |
+
2304,
|
| 149 |
+
1920
|
| 150 |
+
],
|
| 151 |
+
[
|
| 152 |
+
2304,
|
| 153 |
+
2304
|
| 154 |
+
]
|
| 155 |
+
],
|
| 156 |
+
"image_token_index": 151669,
|
| 157 |
+
"model_type": "llava_qwen3_5vit_qwen3",
|
| 158 |
+
"multimodal_projector_bias": true,
|
| 159 |
+
"pad_token_id": 151643,
|
| 160 |
+
"pair_fusion_clip_selection": "",
|
| 161 |
+
"pair_fusion_lr": 0.0001,
|
| 162 |
+
"pair_fusion_motion_cache_format": "",
|
| 163 |
+
"pair_fusion_motion_cache_version": 0,
|
| 164 |
+
"pair_fusion_motion_record_count": 0,
|
| 165 |
+
"pair_fusion_pretraining": "",
|
| 166 |
+
"pair_fusion_pretraining_step": 0,
|
| 167 |
+
"projector_hidden_act": "gelu",
|
| 168 |
+
"shared_image_temporal_refiner": false,
|
| 169 |
+
"temporal_refiner_lr": 0.0001,
|
| 170 |
+
"temporal_refiner_variant": "attention_only",
|
| 171 |
+
"temporal_refiner_window_groups": 4,
|
| 172 |
+
"text_config": {
|
| 173 |
+
"_name_or_path": "/share/m2v_intern_v3/wangjunjie09/model_cache/huggingface/Qwen/Qwen3-8B",
|
| 174 |
+
"architectures": [
|
| 175 |
+
"Qwen3ForCausalLM"
|
| 176 |
+
],
|
| 177 |
+
"attention_bias": false,
|
| 178 |
+
"attention_dropout": 0.0,
|
| 179 |
+
"bos_token_id": 151643,
|
| 180 |
+
"dtype": "bfloat16",
|
| 181 |
+
"eos_token_id": 151645,
|
| 182 |
+
"head_dim": 128,
|
| 183 |
+
"hidden_act": "silu",
|
| 184 |
+
"hidden_size": 4096,
|
| 185 |
+
"initializer_range": 0.02,
|
| 186 |
+
"intermediate_size": 12288,
|
| 187 |
+
"layer_types": [
|
| 188 |
+
"full_attention",
|
| 189 |
+
"full_attention",
|
| 190 |
+
"full_attention",
|
| 191 |
+
"full_attention",
|
| 192 |
+
"full_attention",
|
| 193 |
+
"full_attention",
|
| 194 |
+
"full_attention",
|
| 195 |
+
"full_attention",
|
| 196 |
+
"full_attention",
|
| 197 |
+
"full_attention",
|
| 198 |
+
"full_attention",
|
| 199 |
+
"full_attention",
|
| 200 |
+
"full_attention",
|
| 201 |
+
"full_attention",
|
| 202 |
+
"full_attention",
|
| 203 |
+
"full_attention",
|
| 204 |
+
"full_attention",
|
| 205 |
+
"full_attention",
|
| 206 |
+
"full_attention",
|
| 207 |
+
"full_attention",
|
| 208 |
+
"full_attention",
|
| 209 |
+
"full_attention",
|
| 210 |
+
"full_attention",
|
| 211 |
+
"full_attention",
|
| 212 |
+
"full_attention",
|
| 213 |
+
"full_attention",
|
| 214 |
+
"full_attention",
|
| 215 |
+
"full_attention",
|
| 216 |
+
"full_attention",
|
| 217 |
+
"full_attention",
|
| 218 |
+
"full_attention",
|
| 219 |
+
"full_attention",
|
| 220 |
+
"full_attention",
|
| 221 |
+
"full_attention",
|
| 222 |
+
"full_attention",
|
| 223 |
+
"full_attention"
|
| 224 |
+
],
|
| 225 |
+
"max_position_embeddings": 40960,
|
| 226 |
+
"max_window_layers": 36,
|
| 227 |
+
"model_type": "qwen3",
|
| 228 |
+
"num_attention_heads": 32,
|
| 229 |
+
"num_hidden_layers": 36,
|
| 230 |
+
"num_key_value_heads": 8,
|
| 231 |
+
"pad_token_id": 151643,
|
| 232 |
+
"rms_norm_eps": 1e-06,
|
| 233 |
+
"rope_parameters": {
|
| 234 |
+
"rope_theta": 1000000,
|
| 235 |
+
"rope_type": "default"
|
| 236 |
+
},
|
| 237 |
+
"sliding_window": null,
|
| 238 |
+
"tie_word_embeddings": false,
|
| 239 |
+
"use_cache": false,
|
| 240 |
+
"use_sliding_window": false,
|
| 241 |
+
"vocab_size": 151936
|
| 242 |
+
},
|
| 243 |
+
"tie_word_embeddings": false,
|
| 244 |
+
"train_num_frames": 64,
|
| 245 |
+
"transformers_version": "5.5.4",
|
| 246 |
+
"video_mode": "vtc_compression",
|
| 247 |
+
"video_temporal_refiner": 0,
|
| 248 |
+
"video_token_index": 151670,
|
| 249 |
+
"vision_aspect_ratio": "anyres_max_9",
|
| 250 |
+
"vision_config": {
|
| 251 |
+
"deepstack_visual_indexes": [],
|
| 252 |
+
"depth": 24,
|
| 253 |
+
"dtype": "bfloat16",
|
| 254 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 255 |
+
"hidden_size": 1024,
|
| 256 |
+
"image_size": 384,
|
| 257 |
+
"in_channels": 3,
|
| 258 |
+
"initializer_range": 0.02,
|
| 259 |
+
"intermediate_size": 4096,
|
| 260 |
+
"model_type": "qwen3_5",
|
| 261 |
+
"num_heads": 16,
|
| 262 |
+
"num_position_embeddings": 2304,
|
| 263 |
+
"out_hidden_size": 2048,
|
| 264 |
+
"patch_size": 16,
|
| 265 |
+
"spatial_merge_size": 2,
|
| 266 |
+
"temporal_patch_size": 2,
|
| 267 |
+
"vtc_compressor_type": "vtc_x2",
|
| 268 |
+
"vtc_compressor_variant": "baseline",
|
| 269 |
+
"vtc_feature_fusion_layers": [],
|
| 270 |
+
"vtc_temporal_attention_window": 2
|
| 271 |
+
},
|
| 272 |
+
"vision_feature_layer": -1,
|
| 273 |
+
"vision_feature_select_strategy": "full",
|
| 274 |
+
"vra_layer": 18,
|
| 275 |
+
"vra_loss": 0,
|
| 276 |
+
"vra_scope": "image_video",
|
| 277 |
+
"vra_target_path": "/share/m2v_intern_v3/wangjunjie09/model_cache/huggingface/facebook/dinov3-vitl16-pretrain-lvd1689m",
|
| 278 |
+
"vra_teacher_kind": "dino",
|
| 279 |
+
"vra_teacher_res": 384,
|
| 280 |
+
"vra_video_loss": "direct",
|
| 281 |
+
"vra_weight": 0.5,
|
| 282 |
+
"vra_z_dim": 1024,
|
| 283 |
+
"auto_map": {
|
| 284 |
+
"AutoConfig": "modeling_qwen3_5vit_qwen3.LlavaQwen3_5ViTConfig",
|
| 285 |
+
"AutoModel": "modeling_qwen3_5vit_qwen3.LlavaQwen3_5ViTForConditionalGeneration",
|
| 286 |
+
"AutoModelForCausalLM": "modeling_qwen3_5vit_qwen3.LlavaQwen3_5ViTForConditionalGeneration",
|
| 287 |
+
"AutoModelForImageTextToText": "modeling_qwen3_5vit_qwen3.LlavaQwen3_5ViTForConditionalGeneration"
|
| 288 |
+
}
|
| 289 |
+
}
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/generation_config.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 151643,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
151645,
|
| 6 |
+
151643
|
| 7 |
+
],
|
| 8 |
+
"output_attentions": false,
|
| 9 |
+
"output_hidden_states": false,
|
| 10 |
+
"transformers_version": "5.5.4",
|
| 11 |
+
"use_cache": true
|
| 12 |
+
}
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/model-00001-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:07db5606ac6122fee5feca68f1ee4b6e2affbd920926e977eb4f59642747717d
|
| 3 |
+
size 4905365464
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/model-00002-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8755a5f3c6fde36570a7bf21e72381b81369f67ccfddbb9fcca989a6b4529674
|
| 3 |
+
size 4915962456
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/model-00003-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4cbb5c057187ee03b66a9af18aa2c058882d9f07faad04a14818f37baf75b1f2
|
| 3 |
+
size 4974674312
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/model-00004-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:64a8807aa7cf5ed350eaee3bf57e8c01db0132bd783f295a9ac508406d0b9624
|
| 3 |
+
size 2261019208
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/model.safetensors.index.json
ADDED
|
@@ -0,0 +1,720 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"metadata": {
|
| 3 |
+
"total_parameters": 8528464896,
|
| 4 |
+
"total_size": 17056929792
|
| 5 |
+
},
|
| 6 |
+
"weight_map": {
|
| 7 |
+
"lm_head.weight": "model-00001-of-00004.safetensors",
|
| 8 |
+
"model.image_newline": "model-00001-of-00004.safetensors",
|
| 9 |
+
"model.language_model.embed_tokens.weight": "model-00001-of-00004.safetensors",
|
| 10 |
+
"model.language_model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 11 |
+
"model.language_model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
|
| 12 |
+
"model.language_model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
|
| 13 |
+
"model.language_model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
|
| 14 |
+
"model.language_model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 15 |
+
"model.language_model.layers.0.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 16 |
+
"model.language_model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
|
| 17 |
+
"model.language_model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
|
| 18 |
+
"model.language_model.layers.0.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 19 |
+
"model.language_model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
|
| 20 |
+
"model.language_model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
|
| 21 |
+
"model.language_model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 22 |
+
"model.language_model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
|
| 23 |
+
"model.language_model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
|
| 24 |
+
"model.language_model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
|
| 25 |
+
"model.language_model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 26 |
+
"model.language_model.layers.1.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 27 |
+
"model.language_model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
|
| 28 |
+
"model.language_model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
|
| 29 |
+
"model.language_model.layers.1.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 30 |
+
"model.language_model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
|
| 31 |
+
"model.language_model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
|
| 32 |
+
"model.language_model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 33 |
+
"model.language_model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 34 |
+
"model.language_model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 35 |
+
"model.language_model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 36 |
+
"model.language_model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 37 |
+
"model.language_model.layers.10.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 38 |
+
"model.language_model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 39 |
+
"model.language_model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 40 |
+
"model.language_model.layers.10.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 41 |
+
"model.language_model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 42 |
+
"model.language_model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 43 |
+
"model.language_model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 44 |
+
"model.language_model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 45 |
+
"model.language_model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 46 |
+
"model.language_model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 47 |
+
"model.language_model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 48 |
+
"model.language_model.layers.11.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 49 |
+
"model.language_model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 50 |
+
"model.language_model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 51 |
+
"model.language_model.layers.11.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 52 |
+
"model.language_model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 53 |
+
"model.language_model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 54 |
+
"model.language_model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 55 |
+
"model.language_model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 56 |
+
"model.language_model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 57 |
+
"model.language_model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 58 |
+
"model.language_model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 59 |
+
"model.language_model.layers.12.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 60 |
+
"model.language_model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 61 |
+
"model.language_model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 62 |
+
"model.language_model.layers.12.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 63 |
+
"model.language_model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 64 |
+
"model.language_model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 65 |
+
"model.language_model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 66 |
+
"model.language_model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 67 |
+
"model.language_model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 68 |
+
"model.language_model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 69 |
+
"model.language_model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 70 |
+
"model.language_model.layers.13.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 71 |
+
"model.language_model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 72 |
+
"model.language_model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 73 |
+
"model.language_model.layers.13.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 74 |
+
"model.language_model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 75 |
+
"model.language_model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 76 |
+
"model.language_model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 77 |
+
"model.language_model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 78 |
+
"model.language_model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 79 |
+
"model.language_model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 80 |
+
"model.language_model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 81 |
+
"model.language_model.layers.14.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 82 |
+
"model.language_model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 83 |
+
"model.language_model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 84 |
+
"model.language_model.layers.14.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 85 |
+
"model.language_model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 86 |
+
"model.language_model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 87 |
+
"model.language_model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 88 |
+
"model.language_model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 89 |
+
"model.language_model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 90 |
+
"model.language_model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 91 |
+
"model.language_model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 92 |
+
"model.language_model.layers.15.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 93 |
+
"model.language_model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 94 |
+
"model.language_model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 95 |
+
"model.language_model.layers.15.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 96 |
+
"model.language_model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 97 |
+
"model.language_model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 98 |
+
"model.language_model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 99 |
+
"model.language_model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 100 |
+
"model.language_model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 101 |
+
"model.language_model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 102 |
+
"model.language_model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 103 |
+
"model.language_model.layers.16.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 104 |
+
"model.language_model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 105 |
+
"model.language_model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 106 |
+
"model.language_model.layers.16.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 107 |
+
"model.language_model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 108 |
+
"model.language_model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 109 |
+
"model.language_model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 110 |
+
"model.language_model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 111 |
+
"model.language_model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 112 |
+
"model.language_model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 113 |
+
"model.language_model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 114 |
+
"model.language_model.layers.17.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 115 |
+
"model.language_model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 116 |
+
"model.language_model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 117 |
+
"model.language_model.layers.17.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 118 |
+
"model.language_model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 119 |
+
"model.language_model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 120 |
+
"model.language_model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 121 |
+
"model.language_model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 122 |
+
"model.language_model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 123 |
+
"model.language_model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 124 |
+
"model.language_model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 125 |
+
"model.language_model.layers.18.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 126 |
+
"model.language_model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 127 |
+
"model.language_model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 128 |
+
"model.language_model.layers.18.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 129 |
+
"model.language_model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 130 |
+
"model.language_model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 131 |
+
"model.language_model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 132 |
+
"model.language_model.layers.19.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 133 |
+
"model.language_model.layers.19.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 134 |
+
"model.language_model.layers.19.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 135 |
+
"model.language_model.layers.19.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 136 |
+
"model.language_model.layers.19.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 137 |
+
"model.language_model.layers.19.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 138 |
+
"model.language_model.layers.19.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 139 |
+
"model.language_model.layers.19.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 140 |
+
"model.language_model.layers.19.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 141 |
+
"model.language_model.layers.19.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 142 |
+
"model.language_model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 143 |
+
"model.language_model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
|
| 144 |
+
"model.language_model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
|
| 145 |
+
"model.language_model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
|
| 146 |
+
"model.language_model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 147 |
+
"model.language_model.layers.2.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 148 |
+
"model.language_model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
|
| 149 |
+
"model.language_model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
|
| 150 |
+
"model.language_model.layers.2.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 151 |
+
"model.language_model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
|
| 152 |
+
"model.language_model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
|
| 153 |
+
"model.language_model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 154 |
+
"model.language_model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 155 |
+
"model.language_model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 156 |
+
"model.language_model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 157 |
+
"model.language_model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 158 |
+
"model.language_model.layers.20.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 159 |
+
"model.language_model.layers.20.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 160 |
+
"model.language_model.layers.20.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 161 |
+
"model.language_model.layers.20.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 162 |
+
"model.language_model.layers.20.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 163 |
+
"model.language_model.layers.20.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 164 |
+
"model.language_model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 165 |
+
"model.language_model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 166 |
+
"model.language_model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 167 |
+
"model.language_model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 168 |
+
"model.language_model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 169 |
+
"model.language_model.layers.21.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 170 |
+
"model.language_model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 171 |
+
"model.language_model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 172 |
+
"model.language_model.layers.21.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 173 |
+
"model.language_model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 174 |
+
"model.language_model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 175 |
+
"model.language_model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 176 |
+
"model.language_model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 177 |
+
"model.language_model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 178 |
+
"model.language_model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 179 |
+
"model.language_model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 180 |
+
"model.language_model.layers.22.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 181 |
+
"model.language_model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 182 |
+
"model.language_model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 183 |
+
"model.language_model.layers.22.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 184 |
+
"model.language_model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 185 |
+
"model.language_model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 186 |
+
"model.language_model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 187 |
+
"model.language_model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 188 |
+
"model.language_model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 189 |
+
"model.language_model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 190 |
+
"model.language_model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 191 |
+
"model.language_model.layers.23.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 192 |
+
"model.language_model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 193 |
+
"model.language_model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 194 |
+
"model.language_model.layers.23.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 195 |
+
"model.language_model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 196 |
+
"model.language_model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 197 |
+
"model.language_model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 198 |
+
"model.language_model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 199 |
+
"model.language_model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 200 |
+
"model.language_model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 201 |
+
"model.language_model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 202 |
+
"model.language_model.layers.24.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 203 |
+
"model.language_model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 204 |
+
"model.language_model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 205 |
+
"model.language_model.layers.24.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 206 |
+
"model.language_model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 207 |
+
"model.language_model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 208 |
+
"model.language_model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 209 |
+
"model.language_model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 210 |
+
"model.language_model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 211 |
+
"model.language_model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 212 |
+
"model.language_model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 213 |
+
"model.language_model.layers.25.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 214 |
+
"model.language_model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 215 |
+
"model.language_model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 216 |
+
"model.language_model.layers.25.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 217 |
+
"model.language_model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 218 |
+
"model.language_model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 219 |
+
"model.language_model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 220 |
+
"model.language_model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 221 |
+
"model.language_model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 222 |
+
"model.language_model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 223 |
+
"model.language_model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 224 |
+
"model.language_model.layers.26.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 225 |
+
"model.language_model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 226 |
+
"model.language_model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 227 |
+
"model.language_model.layers.26.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 228 |
+
"model.language_model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 229 |
+
"model.language_model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 230 |
+
"model.language_model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 231 |
+
"model.language_model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 232 |
+
"model.language_model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 233 |
+
"model.language_model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 234 |
+
"model.language_model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 235 |
+
"model.language_model.layers.27.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 236 |
+
"model.language_model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 237 |
+
"model.language_model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 238 |
+
"model.language_model.layers.27.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 239 |
+
"model.language_model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 240 |
+
"model.language_model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 241 |
+
"model.language_model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 242 |
+
"model.language_model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 243 |
+
"model.language_model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 244 |
+
"model.language_model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 245 |
+
"model.language_model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 246 |
+
"model.language_model.layers.28.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 247 |
+
"model.language_model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 248 |
+
"model.language_model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 249 |
+
"model.language_model.layers.28.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 250 |
+
"model.language_model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 251 |
+
"model.language_model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 252 |
+
"model.language_model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 253 |
+
"model.language_model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 254 |
+
"model.language_model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 255 |
+
"model.language_model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 256 |
+
"model.language_model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 257 |
+
"model.language_model.layers.29.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 258 |
+
"model.language_model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 259 |
+
"model.language_model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 260 |
+
"model.language_model.layers.29.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 261 |
+
"model.language_model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 262 |
+
"model.language_model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 263 |
+
"model.language_model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 264 |
+
"model.language_model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
|
| 265 |
+
"model.language_model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
|
| 266 |
+
"model.language_model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
|
| 267 |
+
"model.language_model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 268 |
+
"model.language_model.layers.3.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 269 |
+
"model.language_model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
|
| 270 |
+
"model.language_model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
|
| 271 |
+
"model.language_model.layers.3.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 272 |
+
"model.language_model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
|
| 273 |
+
"model.language_model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
|
| 274 |
+
"model.language_model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 275 |
+
"model.language_model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 276 |
+
"model.language_model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 277 |
+
"model.language_model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 278 |
+
"model.language_model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 279 |
+
"model.language_model.layers.30.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 280 |
+
"model.language_model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 281 |
+
"model.language_model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 282 |
+
"model.language_model.layers.30.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 283 |
+
"model.language_model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
|
| 284 |
+
"model.language_model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
|
| 285 |
+
"model.language_model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 286 |
+
"model.language_model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
|
| 287 |
+
"model.language_model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
|
| 288 |
+
"model.language_model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
|
| 289 |
+
"model.language_model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
|
| 290 |
+
"model.language_model.layers.31.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 291 |
+
"model.language_model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
|
| 292 |
+
"model.language_model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
|
| 293 |
+
"model.language_model.layers.31.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 294 |
+
"model.language_model.layers.31.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
|
| 295 |
+
"model.language_model.layers.31.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
|
| 296 |
+
"model.language_model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 297 |
+
"model.language_model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
|
| 298 |
+
"model.language_model.layers.32.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
|
| 299 |
+
"model.language_model.layers.32.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
|
| 300 |
+
"model.language_model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 301 |
+
"model.language_model.layers.32.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
|
| 302 |
+
"model.language_model.layers.32.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
|
| 303 |
+
"model.language_model.layers.32.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
|
| 304 |
+
"model.language_model.layers.32.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
|
| 305 |
+
"model.language_model.layers.32.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
|
| 306 |
+
"model.language_model.layers.32.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
|
| 307 |
+
"model.language_model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 308 |
+
"model.language_model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
|
| 309 |
+
"model.language_model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
|
| 310 |
+
"model.language_model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
|
| 311 |
+
"model.language_model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 312 |
+
"model.language_model.layers.33.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
|
| 313 |
+
"model.language_model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
|
| 314 |
+
"model.language_model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
|
| 315 |
+
"model.language_model.layers.33.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
|
| 316 |
+
"model.language_model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
|
| 317 |
+
"model.language_model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
|
| 318 |
+
"model.language_model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 319 |
+
"model.language_model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
|
| 320 |
+
"model.language_model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
|
| 321 |
+
"model.language_model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
|
| 322 |
+
"model.language_model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 323 |
+
"model.language_model.layers.34.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
|
| 324 |
+
"model.language_model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
|
| 325 |
+
"model.language_model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
|
| 326 |
+
"model.language_model.layers.34.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
|
| 327 |
+
"model.language_model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
|
| 328 |
+
"model.language_model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
|
| 329 |
+
"model.language_model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 330 |
+
"model.language_model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
|
| 331 |
+
"model.language_model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
|
| 332 |
+
"model.language_model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
|
| 333 |
+
"model.language_model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 334 |
+
"model.language_model.layers.35.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
|
| 335 |
+
"model.language_model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
|
| 336 |
+
"model.language_model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
|
| 337 |
+
"model.language_model.layers.35.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
|
| 338 |
+
"model.language_model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
|
| 339 |
+
"model.language_model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
|
| 340 |
+
"model.language_model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 341 |
+
"model.language_model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
|
| 342 |
+
"model.language_model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
|
| 343 |
+
"model.language_model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
|
| 344 |
+
"model.language_model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 345 |
+
"model.language_model.layers.4.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 346 |
+
"model.language_model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
|
| 347 |
+
"model.language_model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
|
| 348 |
+
"model.language_model.layers.4.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 349 |
+
"model.language_model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
|
| 350 |
+
"model.language_model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
|
| 351 |
+
"model.language_model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 352 |
+
"model.language_model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
|
| 353 |
+
"model.language_model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
|
| 354 |
+
"model.language_model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
|
| 355 |
+
"model.language_model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 356 |
+
"model.language_model.layers.5.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 357 |
+
"model.language_model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
|
| 358 |
+
"model.language_model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
|
| 359 |
+
"model.language_model.layers.5.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 360 |
+
"model.language_model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
|
| 361 |
+
"model.language_model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
|
| 362 |
+
"model.language_model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
|
| 363 |
+
"model.language_model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
|
| 364 |
+
"model.language_model.layers.6.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 365 |
+
"model.language_model.layers.6.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 366 |
+
"model.language_model.layers.6.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 367 |
+
"model.language_model.layers.6.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 368 |
+
"model.language_model.layers.6.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 369 |
+
"model.language_model.layers.6.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 370 |
+
"model.language_model.layers.6.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 371 |
+
"model.language_model.layers.6.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 372 |
+
"model.language_model.layers.6.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 373 |
+
"model.language_model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 374 |
+
"model.language_model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 375 |
+
"model.language_model.layers.7.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 376 |
+
"model.language_model.layers.7.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 377 |
+
"model.language_model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 378 |
+
"model.language_model.layers.7.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 379 |
+
"model.language_model.layers.7.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 380 |
+
"model.language_model.layers.7.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 381 |
+
"model.language_model.layers.7.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 382 |
+
"model.language_model.layers.7.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 383 |
+
"model.language_model.layers.7.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 384 |
+
"model.language_model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 385 |
+
"model.language_model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 386 |
+
"model.language_model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 387 |
+
"model.language_model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 388 |
+
"model.language_model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 389 |
+
"model.language_model.layers.8.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 390 |
+
"model.language_model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 391 |
+
"model.language_model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 392 |
+
"model.language_model.layers.8.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 393 |
+
"model.language_model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 394 |
+
"model.language_model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 395 |
+
"model.language_model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 396 |
+
"model.language_model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
|
| 397 |
+
"model.language_model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
|
| 398 |
+
"model.language_model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
|
| 399 |
+
"model.language_model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
|
| 400 |
+
"model.language_model.layers.9.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 401 |
+
"model.language_model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
|
| 402 |
+
"model.language_model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
|
| 403 |
+
"model.language_model.layers.9.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 404 |
+
"model.language_model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
|
| 405 |
+
"model.language_model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
|
| 406 |
+
"model.language_model.norm.weight": "model-00004-of-00004.safetensors",
|
| 407 |
+
"model.multi_modal_projector.linear_1.bias": "model-00004-of-00004.safetensors",
|
| 408 |
+
"model.multi_modal_projector.linear_1.weight": "model-00004-of-00004.safetensors",
|
| 409 |
+
"model.multi_modal_projector.linear_2.bias": "model-00004-of-00004.safetensors",
|
| 410 |
+
"model.multi_modal_projector.linear_2.weight": "model-00004-of-00004.safetensors",
|
| 411 |
+
"model.vision_tower.vision.blocks.0.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 412 |
+
"model.vision_tower.vision.blocks.0.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 413 |
+
"model.vision_tower.vision.blocks.0.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 414 |
+
"model.vision_tower.vision.blocks.0.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 415 |
+
"model.vision_tower.vision.blocks.0.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 416 |
+
"model.vision_tower.vision.blocks.0.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 417 |
+
"model.vision_tower.vision.blocks.0.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 418 |
+
"model.vision_tower.vision.blocks.0.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 419 |
+
"model.vision_tower.vision.blocks.0.norm1.bias": "model-00004-of-00004.safetensors",
|
| 420 |
+
"model.vision_tower.vision.blocks.0.norm1.weight": "model-00004-of-00004.safetensors",
|
| 421 |
+
"model.vision_tower.vision.blocks.0.norm2.bias": "model-00004-of-00004.safetensors",
|
| 422 |
+
"model.vision_tower.vision.blocks.0.norm2.weight": "model-00004-of-00004.safetensors",
|
| 423 |
+
"model.vision_tower.vision.blocks.1.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 424 |
+
"model.vision_tower.vision.blocks.1.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 425 |
+
"model.vision_tower.vision.blocks.1.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 426 |
+
"model.vision_tower.vision.blocks.1.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 427 |
+
"model.vision_tower.vision.blocks.1.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 428 |
+
"model.vision_tower.vision.blocks.1.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 429 |
+
"model.vision_tower.vision.blocks.1.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 430 |
+
"model.vision_tower.vision.blocks.1.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 431 |
+
"model.vision_tower.vision.blocks.1.norm1.bias": "model-00004-of-00004.safetensors",
|
| 432 |
+
"model.vision_tower.vision.blocks.1.norm1.weight": "model-00004-of-00004.safetensors",
|
| 433 |
+
"model.vision_tower.vision.blocks.1.norm2.bias": "model-00004-of-00004.safetensors",
|
| 434 |
+
"model.vision_tower.vision.blocks.1.norm2.weight": "model-00004-of-00004.safetensors",
|
| 435 |
+
"model.vision_tower.vision.blocks.10.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 436 |
+
"model.vision_tower.vision.blocks.10.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 437 |
+
"model.vision_tower.vision.blocks.10.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 438 |
+
"model.vision_tower.vision.blocks.10.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 439 |
+
"model.vision_tower.vision.blocks.10.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 440 |
+
"model.vision_tower.vision.blocks.10.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 441 |
+
"model.vision_tower.vision.blocks.10.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 442 |
+
"model.vision_tower.vision.blocks.10.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 443 |
+
"model.vision_tower.vision.blocks.10.norm1.bias": "model-00004-of-00004.safetensors",
|
| 444 |
+
"model.vision_tower.vision.blocks.10.norm1.weight": "model-00004-of-00004.safetensors",
|
| 445 |
+
"model.vision_tower.vision.blocks.10.norm2.bias": "model-00004-of-00004.safetensors",
|
| 446 |
+
"model.vision_tower.vision.blocks.10.norm2.weight": "model-00004-of-00004.safetensors",
|
| 447 |
+
"model.vision_tower.vision.blocks.11.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 448 |
+
"model.vision_tower.vision.blocks.11.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 449 |
+
"model.vision_tower.vision.blocks.11.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 450 |
+
"model.vision_tower.vision.blocks.11.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 451 |
+
"model.vision_tower.vision.blocks.11.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 452 |
+
"model.vision_tower.vision.blocks.11.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 453 |
+
"model.vision_tower.vision.blocks.11.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 454 |
+
"model.vision_tower.vision.blocks.11.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 455 |
+
"model.vision_tower.vision.blocks.11.norm1.bias": "model-00004-of-00004.safetensors",
|
| 456 |
+
"model.vision_tower.vision.blocks.11.norm1.weight": "model-00004-of-00004.safetensors",
|
| 457 |
+
"model.vision_tower.vision.blocks.11.norm2.bias": "model-00004-of-00004.safetensors",
|
| 458 |
+
"model.vision_tower.vision.blocks.11.norm2.weight": "model-00004-of-00004.safetensors",
|
| 459 |
+
"model.vision_tower.vision.blocks.12.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 460 |
+
"model.vision_tower.vision.blocks.12.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 461 |
+
"model.vision_tower.vision.blocks.12.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 462 |
+
"model.vision_tower.vision.blocks.12.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 463 |
+
"model.vision_tower.vision.blocks.12.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 464 |
+
"model.vision_tower.vision.blocks.12.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 465 |
+
"model.vision_tower.vision.blocks.12.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 466 |
+
"model.vision_tower.vision.blocks.12.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 467 |
+
"model.vision_tower.vision.blocks.12.norm1.bias": "model-00004-of-00004.safetensors",
|
| 468 |
+
"model.vision_tower.vision.blocks.12.norm1.weight": "model-00004-of-00004.safetensors",
|
| 469 |
+
"model.vision_tower.vision.blocks.12.norm2.bias": "model-00004-of-00004.safetensors",
|
| 470 |
+
"model.vision_tower.vision.blocks.12.norm2.weight": "model-00004-of-00004.safetensors",
|
| 471 |
+
"model.vision_tower.vision.blocks.13.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 472 |
+
"model.vision_tower.vision.blocks.13.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 473 |
+
"model.vision_tower.vision.blocks.13.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 474 |
+
"model.vision_tower.vision.blocks.13.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 475 |
+
"model.vision_tower.vision.blocks.13.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 476 |
+
"model.vision_tower.vision.blocks.13.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 477 |
+
"model.vision_tower.vision.blocks.13.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 478 |
+
"model.vision_tower.vision.blocks.13.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 479 |
+
"model.vision_tower.vision.blocks.13.norm1.bias": "model-00004-of-00004.safetensors",
|
| 480 |
+
"model.vision_tower.vision.blocks.13.norm1.weight": "model-00004-of-00004.safetensors",
|
| 481 |
+
"model.vision_tower.vision.blocks.13.norm2.bias": "model-00004-of-00004.safetensors",
|
| 482 |
+
"model.vision_tower.vision.blocks.13.norm2.weight": "model-00004-of-00004.safetensors",
|
| 483 |
+
"model.vision_tower.vision.blocks.14.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 484 |
+
"model.vision_tower.vision.blocks.14.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 485 |
+
"model.vision_tower.vision.blocks.14.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 486 |
+
"model.vision_tower.vision.blocks.14.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 487 |
+
"model.vision_tower.vision.blocks.14.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 488 |
+
"model.vision_tower.vision.blocks.14.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 489 |
+
"model.vision_tower.vision.blocks.14.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 490 |
+
"model.vision_tower.vision.blocks.14.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 491 |
+
"model.vision_tower.vision.blocks.14.norm1.bias": "model-00004-of-00004.safetensors",
|
| 492 |
+
"model.vision_tower.vision.blocks.14.norm1.weight": "model-00004-of-00004.safetensors",
|
| 493 |
+
"model.vision_tower.vision.blocks.14.norm2.bias": "model-00004-of-00004.safetensors",
|
| 494 |
+
"model.vision_tower.vision.blocks.14.norm2.weight": "model-00004-of-00004.safetensors",
|
| 495 |
+
"model.vision_tower.vision.blocks.15.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 496 |
+
"model.vision_tower.vision.blocks.15.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 497 |
+
"model.vision_tower.vision.blocks.15.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 498 |
+
"model.vision_tower.vision.blocks.15.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 499 |
+
"model.vision_tower.vision.blocks.15.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 500 |
+
"model.vision_tower.vision.blocks.15.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 501 |
+
"model.vision_tower.vision.blocks.15.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 502 |
+
"model.vision_tower.vision.blocks.15.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 503 |
+
"model.vision_tower.vision.blocks.15.norm1.bias": "model-00004-of-00004.safetensors",
|
| 504 |
+
"model.vision_tower.vision.blocks.15.norm1.weight": "model-00004-of-00004.safetensors",
|
| 505 |
+
"model.vision_tower.vision.blocks.15.norm2.bias": "model-00004-of-00004.safetensors",
|
| 506 |
+
"model.vision_tower.vision.blocks.15.norm2.weight": "model-00004-of-00004.safetensors",
|
| 507 |
+
"model.vision_tower.vision.blocks.16.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 508 |
+
"model.vision_tower.vision.blocks.16.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 509 |
+
"model.vision_tower.vision.blocks.16.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 510 |
+
"model.vision_tower.vision.blocks.16.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 511 |
+
"model.vision_tower.vision.blocks.16.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 512 |
+
"model.vision_tower.vision.blocks.16.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 513 |
+
"model.vision_tower.vision.blocks.16.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 514 |
+
"model.vision_tower.vision.blocks.16.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 515 |
+
"model.vision_tower.vision.blocks.16.norm1.bias": "model-00004-of-00004.safetensors",
|
| 516 |
+
"model.vision_tower.vision.blocks.16.norm1.weight": "model-00004-of-00004.safetensors",
|
| 517 |
+
"model.vision_tower.vision.blocks.16.norm2.bias": "model-00004-of-00004.safetensors",
|
| 518 |
+
"model.vision_tower.vision.blocks.16.norm2.weight": "model-00004-of-00004.safetensors",
|
| 519 |
+
"model.vision_tower.vision.blocks.17.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 520 |
+
"model.vision_tower.vision.blocks.17.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 521 |
+
"model.vision_tower.vision.blocks.17.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 522 |
+
"model.vision_tower.vision.blocks.17.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 523 |
+
"model.vision_tower.vision.blocks.17.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 524 |
+
"model.vision_tower.vision.blocks.17.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 525 |
+
"model.vision_tower.vision.blocks.17.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 526 |
+
"model.vision_tower.vision.blocks.17.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 527 |
+
"model.vision_tower.vision.blocks.17.norm1.bias": "model-00004-of-00004.safetensors",
|
| 528 |
+
"model.vision_tower.vision.blocks.17.norm1.weight": "model-00004-of-00004.safetensors",
|
| 529 |
+
"model.vision_tower.vision.blocks.17.norm2.bias": "model-00004-of-00004.safetensors",
|
| 530 |
+
"model.vision_tower.vision.blocks.17.norm2.weight": "model-00004-of-00004.safetensors",
|
| 531 |
+
"model.vision_tower.vision.blocks.18.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 532 |
+
"model.vision_tower.vision.blocks.18.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 533 |
+
"model.vision_tower.vision.blocks.18.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 534 |
+
"model.vision_tower.vision.blocks.18.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 535 |
+
"model.vision_tower.vision.blocks.18.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 536 |
+
"model.vision_tower.vision.blocks.18.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 537 |
+
"model.vision_tower.vision.blocks.18.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 538 |
+
"model.vision_tower.vision.blocks.18.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 539 |
+
"model.vision_tower.vision.blocks.18.norm1.bias": "model-00004-of-00004.safetensors",
|
| 540 |
+
"model.vision_tower.vision.blocks.18.norm1.weight": "model-00004-of-00004.safetensors",
|
| 541 |
+
"model.vision_tower.vision.blocks.18.norm2.bias": "model-00004-of-00004.safetensors",
|
| 542 |
+
"model.vision_tower.vision.blocks.18.norm2.weight": "model-00004-of-00004.safetensors",
|
| 543 |
+
"model.vision_tower.vision.blocks.19.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 544 |
+
"model.vision_tower.vision.blocks.19.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 545 |
+
"model.vision_tower.vision.blocks.19.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 546 |
+
"model.vision_tower.vision.blocks.19.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 547 |
+
"model.vision_tower.vision.blocks.19.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 548 |
+
"model.vision_tower.vision.blocks.19.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 549 |
+
"model.vision_tower.vision.blocks.19.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 550 |
+
"model.vision_tower.vision.blocks.19.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 551 |
+
"model.vision_tower.vision.blocks.19.norm1.bias": "model-00004-of-00004.safetensors",
|
| 552 |
+
"model.vision_tower.vision.blocks.19.norm1.weight": "model-00004-of-00004.safetensors",
|
| 553 |
+
"model.vision_tower.vision.blocks.19.norm2.bias": "model-00004-of-00004.safetensors",
|
| 554 |
+
"model.vision_tower.vision.blocks.19.norm2.weight": "model-00004-of-00004.safetensors",
|
| 555 |
+
"model.vision_tower.vision.blocks.2.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 556 |
+
"model.vision_tower.vision.blocks.2.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 557 |
+
"model.vision_tower.vision.blocks.2.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 558 |
+
"model.vision_tower.vision.blocks.2.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 559 |
+
"model.vision_tower.vision.blocks.2.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 560 |
+
"model.vision_tower.vision.blocks.2.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 561 |
+
"model.vision_tower.vision.blocks.2.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 562 |
+
"model.vision_tower.vision.blocks.2.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 563 |
+
"model.vision_tower.vision.blocks.2.norm1.bias": "model-00004-of-00004.safetensors",
|
| 564 |
+
"model.vision_tower.vision.blocks.2.norm1.weight": "model-00004-of-00004.safetensors",
|
| 565 |
+
"model.vision_tower.vision.blocks.2.norm2.bias": "model-00004-of-00004.safetensors",
|
| 566 |
+
"model.vision_tower.vision.blocks.2.norm2.weight": "model-00004-of-00004.safetensors",
|
| 567 |
+
"model.vision_tower.vision.blocks.20.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 568 |
+
"model.vision_tower.vision.blocks.20.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 569 |
+
"model.vision_tower.vision.blocks.20.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 570 |
+
"model.vision_tower.vision.blocks.20.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 571 |
+
"model.vision_tower.vision.blocks.20.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 572 |
+
"model.vision_tower.vision.blocks.20.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 573 |
+
"model.vision_tower.vision.blocks.20.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 574 |
+
"model.vision_tower.vision.blocks.20.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 575 |
+
"model.vision_tower.vision.blocks.20.norm1.bias": "model-00004-of-00004.safetensors",
|
| 576 |
+
"model.vision_tower.vision.blocks.20.norm1.weight": "model-00004-of-00004.safetensors",
|
| 577 |
+
"model.vision_tower.vision.blocks.20.norm2.bias": "model-00004-of-00004.safetensors",
|
| 578 |
+
"model.vision_tower.vision.blocks.20.norm2.weight": "model-00004-of-00004.safetensors",
|
| 579 |
+
"model.vision_tower.vision.blocks.21.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 580 |
+
"model.vision_tower.vision.blocks.21.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 581 |
+
"model.vision_tower.vision.blocks.21.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 582 |
+
"model.vision_tower.vision.blocks.21.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 583 |
+
"model.vision_tower.vision.blocks.21.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 584 |
+
"model.vision_tower.vision.blocks.21.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 585 |
+
"model.vision_tower.vision.blocks.21.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 586 |
+
"model.vision_tower.vision.blocks.21.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 587 |
+
"model.vision_tower.vision.blocks.21.norm1.bias": "model-00004-of-00004.safetensors",
|
| 588 |
+
"model.vision_tower.vision.blocks.21.norm1.weight": "model-00004-of-00004.safetensors",
|
| 589 |
+
"model.vision_tower.vision.blocks.21.norm2.bias": "model-00004-of-00004.safetensors",
|
| 590 |
+
"model.vision_tower.vision.blocks.21.norm2.weight": "model-00004-of-00004.safetensors",
|
| 591 |
+
"model.vision_tower.vision.blocks.22.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 592 |
+
"model.vision_tower.vision.blocks.22.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 593 |
+
"model.vision_tower.vision.blocks.22.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 594 |
+
"model.vision_tower.vision.blocks.22.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 595 |
+
"model.vision_tower.vision.blocks.22.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 596 |
+
"model.vision_tower.vision.blocks.22.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 597 |
+
"model.vision_tower.vision.blocks.22.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 598 |
+
"model.vision_tower.vision.blocks.22.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 599 |
+
"model.vision_tower.vision.blocks.22.norm1.bias": "model-00004-of-00004.safetensors",
|
| 600 |
+
"model.vision_tower.vision.blocks.22.norm1.weight": "model-00004-of-00004.safetensors",
|
| 601 |
+
"model.vision_tower.vision.blocks.22.norm2.bias": "model-00004-of-00004.safetensors",
|
| 602 |
+
"model.vision_tower.vision.blocks.22.norm2.weight": "model-00004-of-00004.safetensors",
|
| 603 |
+
"model.vision_tower.vision.blocks.23.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 604 |
+
"model.vision_tower.vision.blocks.23.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 605 |
+
"model.vision_tower.vision.blocks.23.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 606 |
+
"model.vision_tower.vision.blocks.23.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 607 |
+
"model.vision_tower.vision.blocks.23.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 608 |
+
"model.vision_tower.vision.blocks.23.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 609 |
+
"model.vision_tower.vision.blocks.23.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 610 |
+
"model.vision_tower.vision.blocks.23.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 611 |
+
"model.vision_tower.vision.blocks.23.norm1.bias": "model-00004-of-00004.safetensors",
|
| 612 |
+
"model.vision_tower.vision.blocks.23.norm1.weight": "model-00004-of-00004.safetensors",
|
| 613 |
+
"model.vision_tower.vision.blocks.23.norm2.bias": "model-00004-of-00004.safetensors",
|
| 614 |
+
"model.vision_tower.vision.blocks.23.norm2.weight": "model-00004-of-00004.safetensors",
|
| 615 |
+
"model.vision_tower.vision.blocks.3.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 616 |
+
"model.vision_tower.vision.blocks.3.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 617 |
+
"model.vision_tower.vision.blocks.3.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 618 |
+
"model.vision_tower.vision.blocks.3.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 619 |
+
"model.vision_tower.vision.blocks.3.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 620 |
+
"model.vision_tower.vision.blocks.3.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 621 |
+
"model.vision_tower.vision.blocks.3.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 622 |
+
"model.vision_tower.vision.blocks.3.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 623 |
+
"model.vision_tower.vision.blocks.3.norm1.bias": "model-00004-of-00004.safetensors",
|
| 624 |
+
"model.vision_tower.vision.blocks.3.norm1.weight": "model-00004-of-00004.safetensors",
|
| 625 |
+
"model.vision_tower.vision.blocks.3.norm2.bias": "model-00004-of-00004.safetensors",
|
| 626 |
+
"model.vision_tower.vision.blocks.3.norm2.weight": "model-00004-of-00004.safetensors",
|
| 627 |
+
"model.vision_tower.vision.blocks.4.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 628 |
+
"model.vision_tower.vision.blocks.4.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 629 |
+
"model.vision_tower.vision.blocks.4.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 630 |
+
"model.vision_tower.vision.blocks.4.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 631 |
+
"model.vision_tower.vision.blocks.4.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 632 |
+
"model.vision_tower.vision.blocks.4.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 633 |
+
"model.vision_tower.vision.blocks.4.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 634 |
+
"model.vision_tower.vision.blocks.4.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 635 |
+
"model.vision_tower.vision.blocks.4.norm1.bias": "model-00004-of-00004.safetensors",
|
| 636 |
+
"model.vision_tower.vision.blocks.4.norm1.weight": "model-00004-of-00004.safetensors",
|
| 637 |
+
"model.vision_tower.vision.blocks.4.norm2.bias": "model-00004-of-00004.safetensors",
|
| 638 |
+
"model.vision_tower.vision.blocks.4.norm2.weight": "model-00004-of-00004.safetensors",
|
| 639 |
+
"model.vision_tower.vision.blocks.5.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 640 |
+
"model.vision_tower.vision.blocks.5.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 641 |
+
"model.vision_tower.vision.blocks.5.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 642 |
+
"model.vision_tower.vision.blocks.5.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 643 |
+
"model.vision_tower.vision.blocks.5.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 644 |
+
"model.vision_tower.vision.blocks.5.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 645 |
+
"model.vision_tower.vision.blocks.5.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 646 |
+
"model.vision_tower.vision.blocks.5.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 647 |
+
"model.vision_tower.vision.blocks.5.norm1.bias": "model-00004-of-00004.safetensors",
|
| 648 |
+
"model.vision_tower.vision.blocks.5.norm1.weight": "model-00004-of-00004.safetensors",
|
| 649 |
+
"model.vision_tower.vision.blocks.5.norm2.bias": "model-00004-of-00004.safetensors",
|
| 650 |
+
"model.vision_tower.vision.blocks.5.norm2.weight": "model-00004-of-00004.safetensors",
|
| 651 |
+
"model.vision_tower.vision.blocks.6.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 652 |
+
"model.vision_tower.vision.blocks.6.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 653 |
+
"model.vision_tower.vision.blocks.6.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 654 |
+
"model.vision_tower.vision.blocks.6.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 655 |
+
"model.vision_tower.vision.blocks.6.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 656 |
+
"model.vision_tower.vision.blocks.6.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 657 |
+
"model.vision_tower.vision.blocks.6.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 658 |
+
"model.vision_tower.vision.blocks.6.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 659 |
+
"model.vision_tower.vision.blocks.6.norm1.bias": "model-00004-of-00004.safetensors",
|
| 660 |
+
"model.vision_tower.vision.blocks.6.norm1.weight": "model-00004-of-00004.safetensors",
|
| 661 |
+
"model.vision_tower.vision.blocks.6.norm2.bias": "model-00004-of-00004.safetensors",
|
| 662 |
+
"model.vision_tower.vision.blocks.6.norm2.weight": "model-00004-of-00004.safetensors",
|
| 663 |
+
"model.vision_tower.vision.blocks.7.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 664 |
+
"model.vision_tower.vision.blocks.7.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 665 |
+
"model.vision_tower.vision.blocks.7.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 666 |
+
"model.vision_tower.vision.blocks.7.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 667 |
+
"model.vision_tower.vision.blocks.7.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 668 |
+
"model.vision_tower.vision.blocks.7.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 669 |
+
"model.vision_tower.vision.blocks.7.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 670 |
+
"model.vision_tower.vision.blocks.7.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 671 |
+
"model.vision_tower.vision.blocks.7.norm1.bias": "model-00004-of-00004.safetensors",
|
| 672 |
+
"model.vision_tower.vision.blocks.7.norm1.weight": "model-00004-of-00004.safetensors",
|
| 673 |
+
"model.vision_tower.vision.blocks.7.norm2.bias": "model-00004-of-00004.safetensors",
|
| 674 |
+
"model.vision_tower.vision.blocks.7.norm2.weight": "model-00004-of-00004.safetensors",
|
| 675 |
+
"model.vision_tower.vision.blocks.8.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 676 |
+
"model.vision_tower.vision.blocks.8.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 677 |
+
"model.vision_tower.vision.blocks.8.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 678 |
+
"model.vision_tower.vision.blocks.8.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 679 |
+
"model.vision_tower.vision.blocks.8.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 680 |
+
"model.vision_tower.vision.blocks.8.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 681 |
+
"model.vision_tower.vision.blocks.8.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 682 |
+
"model.vision_tower.vision.blocks.8.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 683 |
+
"model.vision_tower.vision.blocks.8.norm1.bias": "model-00004-of-00004.safetensors",
|
| 684 |
+
"model.vision_tower.vision.blocks.8.norm1.weight": "model-00004-of-00004.safetensors",
|
| 685 |
+
"model.vision_tower.vision.blocks.8.norm2.bias": "model-00004-of-00004.safetensors",
|
| 686 |
+
"model.vision_tower.vision.blocks.8.norm2.weight": "model-00004-of-00004.safetensors",
|
| 687 |
+
"model.vision_tower.vision.blocks.9.attn.proj.bias": "model-00004-of-00004.safetensors",
|
| 688 |
+
"model.vision_tower.vision.blocks.9.attn.proj.weight": "model-00004-of-00004.safetensors",
|
| 689 |
+
"model.vision_tower.vision.blocks.9.attn.qkv.bias": "model-00004-of-00004.safetensors",
|
| 690 |
+
"model.vision_tower.vision.blocks.9.attn.qkv.weight": "model-00004-of-00004.safetensors",
|
| 691 |
+
"model.vision_tower.vision.blocks.9.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
|
| 692 |
+
"model.vision_tower.vision.blocks.9.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
|
| 693 |
+
"model.vision_tower.vision.blocks.9.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
|
| 694 |
+
"model.vision_tower.vision.blocks.9.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
|
| 695 |
+
"model.vision_tower.vision.blocks.9.norm1.bias": "model-00004-of-00004.safetensors",
|
| 696 |
+
"model.vision_tower.vision.blocks.9.norm1.weight": "model-00004-of-00004.safetensors",
|
| 697 |
+
"model.vision_tower.vision.blocks.9.norm2.bias": "model-00004-of-00004.safetensors",
|
| 698 |
+
"model.vision_tower.vision.blocks.9.norm2.weight": "model-00004-of-00004.safetensors",
|
| 699 |
+
"model.vision_tower.vision.final_layernorm.bias": "model-00004-of-00004.safetensors",
|
| 700 |
+
"model.vision_tower.vision.final_layernorm.weight": "model-00004-of-00004.safetensors",
|
| 701 |
+
"model.vision_tower.vision.patch_embed.proj.bias": "model-00004-of-00004.safetensors",
|
| 702 |
+
"model.vision_tower.vision.patch_embed.proj.weight": "model-00004-of-00004.safetensors",
|
| 703 |
+
"model.vision_tower.vision.pos_embed.weight": "model-00004-of-00004.safetensors",
|
| 704 |
+
"model.vision_tower.vtc_compressor.downsample.bias": "model-00004-of-00004.safetensors",
|
| 705 |
+
"model.vision_tower.vtc_compressor.downsample.weight": "model-00004-of-00004.safetensors",
|
| 706 |
+
"model.vision_tower.vtc_compressor.spatial_attention.norm.bias": "model-00004-of-00004.safetensors",
|
| 707 |
+
"model.vision_tower.vtc_compressor.spatial_attention.norm.weight": "model-00004-of-00004.safetensors",
|
| 708 |
+
"model.vision_tower.vtc_compressor.spatial_attention.proj.bias": "model-00004-of-00004.safetensors",
|
| 709 |
+
"model.vision_tower.vtc_compressor.spatial_attention.proj.weight": "model-00004-of-00004.safetensors",
|
| 710 |
+
"model.vision_tower.vtc_compressor.spatial_attention.qkv.bias": "model-00004-of-00004.safetensors",
|
| 711 |
+
"model.vision_tower.vtc_compressor.spatial_attention.qkv.weight": "model-00004-of-00004.safetensors",
|
| 712 |
+
"model.vision_tower.vtc_compressor.temporal_attention.norm.bias": "model-00004-of-00004.safetensors",
|
| 713 |
+
"model.vision_tower.vtc_compressor.temporal_attention.norm.weight": "model-00004-of-00004.safetensors",
|
| 714 |
+
"model.vision_tower.vtc_compressor.temporal_attention.proj.bias": "model-00004-of-00004.safetensors",
|
| 715 |
+
"model.vision_tower.vtc_compressor.temporal_attention.proj.weight": "model-00004-of-00004.safetensors",
|
| 716 |
+
"model.vision_tower.vtc_compressor.temporal_attention.qkv.bias": "model-00004-of-00004.safetensors",
|
| 717 |
+
"model.vision_tower.vtc_compressor.temporal_attention.qkv.weight": "model-00004-of-00004.safetensors",
|
| 718 |
+
"model.vision_tower.vtc_compressor.temporal_position": "model-00004-of-00004.safetensors"
|
| 719 |
+
}
|
| 720 |
+
}
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/modeling_qwen3_5vit_qwen3.py
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/processing_qwen3_5vit_qwen3.py
ADDED
|
@@ -0,0 +1,302 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""processing_qwen3_5vit_qwen3.py — LlavaOnevision processor with optional
|
| 2 |
+
video temporal-merge (tmerge) token accounting.
|
| 3 |
+
|
| 4 |
+
Why a subclass: at eval the model is loaded via HF (lmms-eval `llava_hf` →
|
| 5 |
+
`AutoProcessor.from_pretrained(ckpt, trust_remote_code=True)`), NOT the swift
|
| 6 |
+
template. The `LlavaOnevisionProcessor.__call__` expands `<video>` by
|
| 7 |
+
`num_frames*144+1`. Under `video_temporal_merge=1` the model emits
|
| 8 |
+
`(num_frames//2)*144+1` features (grid_t=T/2 temporal groups), so the default
|
| 9 |
+
count over-expands → `masked_scatter` token/feature mismatch crash.
|
| 10 |
+
|
| 11 |
+
This subclass is byte-identical to the upstream processor EXCEPT the single video
|
| 12 |
+
token-count line, which routes through `qwen3_5_video_token_count` — the same
|
| 13 |
+
SoT helper the swift training template uses. The mllm_eval adapter
|
| 14 |
+
(`mllm_eval/checkpoint/__init__.py::qwen3_5vit`) declares this file as
|
| 15 |
+
`processor_file`, so prep cp's it into the ckpt and injects `auto_map.AutoProcessor`.
|
| 16 |
+
|
| 17 |
+
`video_temporal_merge` is a persisted init kwarg (lands in processor_config.json),
|
| 18 |
+
set by the plugin `get_processor` from env/config so eval inherits it with no env.
|
| 19 |
+
"""
|
| 20 |
+
|
| 21 |
+
import math
|
| 22 |
+
|
| 23 |
+
import numpy as np
|
| 24 |
+
|
| 25 |
+
from transformers.feature_extraction_utils import BatchFeature
|
| 26 |
+
from transformers.image_processing_utils import select_best_resolution
|
| 27 |
+
from transformers.image_utils import ImageInput, get_image_size, to_numpy_array
|
| 28 |
+
from transformers.processing_utils import Unpack
|
| 29 |
+
from transformers.tokenization_utils_base import PreTokenizedInput, TextInput
|
| 30 |
+
from transformers.video_utils import VideoInput
|
| 31 |
+
from transformers.models.llava_onevision.processing_llava_onevision import (
|
| 32 |
+
LlavaOnevisionProcessor,
|
| 33 |
+
LlavaOnevisionProcessorKwargs,
|
| 34 |
+
)
|
| 35 |
+
|
| 36 |
+
from typing import Optional
|
| 37 |
+
|
| 38 |
+
# Flat-name sibling import: works both in-repo (model_plugin adds video_mllm/model/ to
|
| 39 |
+
# sys.path) and at eval (HF dynamic_module loads modeling_qwen3_5vit_qwen3.py as a
|
| 40 |
+
# ckpt-dir sibling — the qwen3_5vit adapter ships it as modeling_file).
|
| 41 |
+
from modeling_qwen3_5vit_qwen3 import (
|
| 42 |
+
build_faithful_video_token_string,
|
| 43 |
+
vtc_compression_factor,
|
| 44 |
+
vtc_temporal_patch_size,
|
| 45 |
+
qwen3_5_video_token_count,
|
| 46 |
+
video_is_vtc_compression,
|
| 47 |
+
video_is_spatial_catp,
|
| 48 |
+
video_is_image_mrope,
|
| 49 |
+
video_is_pair_fusion,
|
| 50 |
+
video_is_temporal_compression,
|
| 51 |
+
video_is_temporal_merge,
|
| 52 |
+
video_mrope_active,
|
| 53 |
+
)
|
| 54 |
+
|
| 55 |
+
|
| 56 |
+
class LlavaQwen3_5ViTProcessor(LlavaOnevisionProcessor):
|
| 57 |
+
"""LlavaOnevisionProcessor whose `<video>` expansion is tmerge-aware."""
|
| 58 |
+
|
| 59 |
+
def __init__(
|
| 60 |
+
self,
|
| 61 |
+
image_processor=None,
|
| 62 |
+
tokenizer=None,
|
| 63 |
+
video_processor=None,
|
| 64 |
+
num_image_tokens=None,
|
| 65 |
+
vision_feature_select_strategy=None,
|
| 66 |
+
chat_template=None,
|
| 67 |
+
image_token="<image>",
|
| 68 |
+
video_token="<video>",
|
| 69 |
+
vision_aspect_ratio="anyres_max_9",
|
| 70 |
+
video_mode: Optional[str] = None,
|
| 71 |
+
vtc_compression_factor: int = 2,
|
| 72 |
+
vtc_temporal_patch_size: int = 1,
|
| 73 |
+
# Legacy int flags kept in the signature ONLY so a pre-2026-06-22 processor_config.json
|
| 74 |
+
# still reaches __init__ on reload (ProcessorMixin filters keys NOT in the signature BEFORE
|
| 75 |
+
# calling __init__). Folded into video_mode and NEVER set as attributes -> to_dict (which
|
| 76 |
+
# dumps __dict__) serializes ONLY video_mode (single SoT; old ckpts migrate on re-save).
|
| 77 |
+
video_temporal_merge: int = 0,
|
| 78 |
+
image_mrope: int = 0,
|
| 79 |
+
video_mrope: int = 0,
|
| 80 |
+
**kwargs,
|
| 81 |
+
):
|
| 82 |
+
# The parent's sub-processors AND scalar kwargs MUST be explicit in THIS signature -- NOT
|
| 83 |
+
# swallowed by a bare (*args, **kwargs). ProcessorMixin serialization enumerates what to
|
| 84 |
+
# persist via get_attributes() = inspect.signature(cls.__init__).parameters; a (*args,
|
| 85 |
+
# **kwargs) signature returns [] -> save_pretrained writes neither the sub-processor files
|
| 86 |
+
# nor video_mode into the ckpt. So video_mode is an explicit, serialized param here.
|
| 87 |
+
# Backward-compat: an old processor_config.json carries the legacy int flags (received via
|
| 88 |
+
# the explicit params above so reload reaches here) and no video_mode. Fold them into
|
| 89 |
+
# video_mode; do NOT set them as attributes, so to_dict writes only video_mode (single SoT;
|
| 90 |
+
# old ckpts migrate on re-save). New ckpts pass video_mode directly (fold skipped).
|
| 91 |
+
if video_mode is None:
|
| 92 |
+
if int(image_mrope or 0):
|
| 93 |
+
video_mode = "image_mrope"
|
| 94 |
+
elif int(video_mrope or 0):
|
| 95 |
+
video_mode = "video_mrope"
|
| 96 |
+
elif int(video_temporal_merge or 0):
|
| 97 |
+
video_mode = "temporal_merge"
|
| 98 |
+
self.video_mode = video_mode
|
| 99 |
+
self.vtc_compression_factor = vtc_compression_factor
|
| 100 |
+
self.vtc_temporal_patch_size = vtc_temporal_patch_size
|
| 101 |
+
super().__init__(
|
| 102 |
+
image_processor=image_processor,
|
| 103 |
+
tokenizer=tokenizer,
|
| 104 |
+
video_processor=video_processor,
|
| 105 |
+
num_image_tokens=num_image_tokens,
|
| 106 |
+
vision_feature_select_strategy=vision_feature_select_strategy,
|
| 107 |
+
chat_template=chat_template,
|
| 108 |
+
image_token=image_token,
|
| 109 |
+
video_token=video_token,
|
| 110 |
+
vision_aspect_ratio=vision_aspect_ratio,
|
| 111 |
+
**kwargs,
|
| 112 |
+
)
|
| 113 |
+
|
| 114 |
+
def __call__(
|
| 115 |
+
self,
|
| 116 |
+
images: ImageInput = None,
|
| 117 |
+
text: "TextInput | PreTokenizedInput | list[TextInput] | list[PreTokenizedInput]" = None,
|
| 118 |
+
videos: VideoInput = None,
|
| 119 |
+
video_fps: float = None,
|
| 120 |
+
video_sample_indices=None,
|
| 121 |
+
**kwargs: Unpack[LlavaOnevisionProcessorKwargs],
|
| 122 |
+
) -> BatchFeature:
|
| 123 |
+
# ── verbatim from LlavaOnevisionProcessor.__call__ (transformers) ──
|
| 124 |
+
output_kwargs = self._merge_kwargs(
|
| 125 |
+
LlavaOnevisionProcessorKwargs,
|
| 126 |
+
tokenizer_init_kwargs=self.tokenizer.init_kwargs,
|
| 127 |
+
**kwargs,
|
| 128 |
+
)
|
| 129 |
+
|
| 130 |
+
if isinstance(text, str):
|
| 131 |
+
text = [text]
|
| 132 |
+
elif not isinstance(text, list) and not isinstance(text[0], str):
|
| 133 |
+
raise TypeError("Invalid input text. Please provide a string, or a list of strings")
|
| 134 |
+
|
| 135 |
+
image_inputs = video_inputs = {}
|
| 136 |
+
|
| 137 |
+
if images is not None:
|
| 138 |
+
image_inputs = self.image_processor(images, **output_kwargs["images_kwargs"])
|
| 139 |
+
|
| 140 |
+
batch_num_images = iter(image_inputs["batch_num_images"])
|
| 141 |
+
image_sizes = iter(image_inputs["image_sizes"])
|
| 142 |
+
height, width = get_image_size(
|
| 143 |
+
to_numpy_array(image_inputs["pixel_values"][0][0]),
|
| 144 |
+
channel_dim=output_kwargs["images_kwargs"].get("data_format"),
|
| 145 |
+
)
|
| 146 |
+
text, num_image_tokens = self._expand_image_tokens(
|
| 147 |
+
text, image_sizes, height, width, self.image_token, batch_num_images
|
| 148 |
+
)
|
| 149 |
+
|
| 150 |
+
if videos is not None:
|
| 151 |
+
video_inputs = self.video_processor(videos, **output_kwargs["videos_kwargs"])
|
| 152 |
+
|
| 153 |
+
one_video = video_inputs.get("pixel_values_videos")[0]
|
| 154 |
+
if isinstance(video_inputs.get("pixel_values_videos")[0], (list, tuple)):
|
| 155 |
+
one_video = np.array(one_video)
|
| 156 |
+
else:
|
| 157 |
+
one_video = to_numpy_array(one_video)
|
| 158 |
+
num_frames = one_video.shape[0] # frame dim is always after batch dim
|
| 159 |
+
# ── ONLY deviation from upstream __call__: the video token count.
|
| 160 |
+
# Upstream derives it inline as `get_image_size(one_video[0])` (unused
|
| 161 |
+
# for count) + `num_frames * ceil(sqrt(num_image_tokens)/2)**2 + 1`.
|
| 162 |
+
# We drop the get_image_size line ON PURPOSE — the SoT helper hardcodes
|
| 163 |
+
# 144 tokens/readout-unit (= ceil(24/2)**2 for Qwen3.5 384/16 grid) and
|
| 164 |
+
# folds in the grid_t=num_frames//2 tubelet grouping when merge is on.
|
| 165 |
+
# Do NOT "restore" get_image_size when re-syncing against the default.
|
| 166 |
+
if video_mrope_active(self):
|
| 167 |
+
# FAIL-FAST: the faithful eval path builds ONE timestamp stream from a single video's
|
| 168 |
+
# (fps, sampled_indices) and replaces ONE <video> placeholder. A multi-video sample
|
| 169 |
+
# or batched (B>1) call would mis-apply the first video's timestamps to the rest.
|
| 170 |
+
# lmms-eval llava_hf runs generate at bs=1 single-video, so this never fires; it
|
| 171 |
+
# guards a future multi-video / batched entry point from silent mis-scoring.
|
| 172 |
+
n_video_ph = sum(sample.count(self.video_token) for sample in text)
|
| 173 |
+
if n_video_ph != 1:
|
| 174 |
+
raise RuntimeError(
|
| 175 |
+
f"LlavaQwen3_5ViTProcessor: faithful video_mrope eval supports exactly ONE "
|
| 176 |
+
f"<video> per call (single video, batch_size=1); found {n_video_ph}. "
|
| 177 |
+
f"Multi-video / batched faithful eval is unsupported."
|
| 178 |
+
)
|
| 179 |
+
if video_fps is None:
|
| 180 |
+
# FAIL-FAST (not a silent blob fallback): video_mrope=1 means the ckpt was trained
|
| 181 |
+
# on the faithful per-frame timestamp stream, so evaluating it on the contiguous
|
| 182 |
+
# blob is a train/eval mismatch that produces silently-wrong numbers. Only the
|
| 183 |
+
# lmms-eval simple (--force_simple) path threads fps today; the chat / VLMEvalKit
|
| 184 |
+
# video paths do not. Current 6-bench video tasks all use --force_simple, so this
|
| 185 |
+
# never fires; it guards a future entry-point swap from silently degrading.
|
| 186 |
+
# (Image-only eval never reaches here — `videos is None` skips this whole block.)
|
| 187 |
+
raise RuntimeError(
|
| 188 |
+
"LlavaQwen3_5ViTProcessor: video_mrope=1 but no video_fps/video_sample_indices "
|
| 189 |
+
"were provided — cannot build the faithful per-frame timestamp stream. Route "
|
| 190 |
+
"video eval through the lmms-eval simple (--force_simple) path, which threads fps."
|
| 191 |
+
)
|
| 192 |
+
# Faithful video-mRoPE (video_mrope=1): build the per-frame timestamp + vision-marker
|
| 193 |
+
# stream via the SAME shared SoT the training template _encode uses, so eval and
|
| 194 |
+
# train produce byte-identical video token ids (zero drift). Needs fps + sampled
|
| 195 |
+
# frame indices (threaded from the eval video loader). One video per eval request.
|
| 196 |
+
faithful_str = build_faithful_video_token_string(
|
| 197 |
+
num_frames, video_fps, video_sample_indices,
|
| 198 |
+
video_is_temporal_merge(self), video_token=self.video_token,
|
| 199 |
+
)
|
| 200 |
+
text = [sample.replace(self.video_token, faithful_str, 1) for sample in text]
|
| 201 |
+
else:
|
| 202 |
+
# Default / non-mrope: byte-identical contiguous blob.
|
| 203 |
+
num_video_tokens = qwen3_5_video_token_count(
|
| 204 |
+
num_frames,
|
| 205 |
+
temporal_merge=video_is_temporal_merge(self),
|
| 206 |
+
temporal_compression=video_is_temporal_compression(self),
|
| 207 |
+
pair_fusion=video_is_pair_fusion(self),
|
| 208 |
+
vtc_compression=video_is_vtc_compression(self),
|
| 209 |
+
spatial_catp=video_is_spatial_catp(self),
|
| 210 |
+
vtc_compression_factor=vtc_compression_factor(self),
|
| 211 |
+
vtc_temporal_patch_size=vtc_temporal_patch_size(self),
|
| 212 |
+
)
|
| 213 |
+
text = [sample.replace(self.video_token, self.video_token * num_video_tokens) for sample in text]
|
| 214 |
+
|
| 215 |
+
return_tensors = output_kwargs["text_kwargs"].pop("return_tensors", None)
|
| 216 |
+
return_mm_token_type_ids = output_kwargs["text_kwargs"].pop("return_mm_token_type_ids", None)
|
| 217 |
+
text_inputs = self.tokenizer(text, **output_kwargs["text_kwargs"])
|
| 218 |
+
self._check_special_mm_tokens(text, text_inputs, modalities=["image"])
|
| 219 |
+
|
| 220 |
+
if return_mm_token_type_ids:
|
| 221 |
+
# Byte-identical to LlavaOnevisionProcessor.__call__ (transformers 5.5.4):
|
| 222 |
+
# delegate to ProcessorMixin.create_mm_token_type_ids, which marks image=1 / video=2
|
| 223 |
+
# / audio=3 over self.image_ids/video_ids/audio_ids (set by the base __init__, so
|
| 224 |
+
# present on the fresh-build AND the __class__-retype reload path). An earlier inline
|
| 225 |
+
# copy here came from an OLDER transformers that lacked this method and marked
|
| 226 |
+
# image-only (mis-typing video tokens); 5.5.4 HAS the method (processing_utils.py) and
|
| 227 |
+
# the default path calls it, so tracking the default keeps the <video> token COUNT the only intentional
|
| 228 |
+
# deviation in this __call__. (This path is dormant: return_mm_token_type_ids defaults
|
| 229 |
+
# False for LlavaOnevision.)
|
| 230 |
+
text_inputs["mm_token_type_ids"] = self.create_mm_token_type_ids(text_inputs["input_ids"])
|
| 231 |
+
return BatchFeature(data={**text_inputs, **image_inputs, **video_inputs}, tensor_type=return_tensors)
|
| 232 |
+
|
| 233 |
+
def _get_number_of_features(self, orig_height: int, orig_width: int, height: int, width: int) -> int:
|
| 234 |
+
"""image_mrope=1: drop the per-row image_newline from the count (the model calls
|
| 235 |
+
pack_image_features with image_newline=None), so the image token count is
|
| 236 |
+
base + unpadded (clean H x W grid). Default count = base + unpadded + newline; here we
|
| 237 |
+
subtract exactly `newline_features`. flag 0 (default / video_mrope) → byte-identical to parent.
|
| 238 |
+
Used by BOTH the swift training _encode and eval __call__/_expand_image_tokens, so the
|
| 239 |
+
model's pack feature count and the input_ids placeholder count stay matched."""
|
| 240 |
+
n = super()._get_number_of_features(orig_height, orig_width, height, width)
|
| 241 |
+
if video_is_image_mrope(self):
|
| 242 |
+
# Guard the one config that would desync the position-id block sum from the emitted
|
| 243 |
+
# token count: vision_feature_select_strategy="default" makes _expand_image_tokens
|
| 244 |
+
# subtract 1 (a CLS-token drop), but image_mrope_grid_blocks emits the full base+
|
| 245 |
+
# unpadded grid (no CLS to drop). Qwen3.5 ViT has NO CLS token, so the strategy is
|
| 246 |
+
# always 'full'/None here; fail loud instead of silently misaligning if that ever changes.
|
| 247 |
+
if getattr(self, "vision_feature_select_strategy", None) == "default":
|
| 248 |
+
raise ValueError(
|
| 249 |
+
"image_mrope is incompatible with vision_feature_select_strategy='default' "
|
| 250 |
+
"(no CLS token in the clean H x W grid; the -1 would desync the position-id "
|
| 251 |
+
"block sum from the token count). Use 'full' / None."
|
| 252 |
+
)
|
| 253 |
+
image_grid_pinpoints = self.image_processor.image_grid_pinpoints
|
| 254 |
+
height_best, width_best = select_best_resolution([orig_height, orig_width], image_grid_pinpoints)
|
| 255 |
+
scale_height, scale_width = height_best // height, width_best // width
|
| 256 |
+
patches = int(math.sqrt(self.num_image_tokens))
|
| 257 |
+
_, newline_features = self._get_unpadded_features(
|
| 258 |
+
orig_height, orig_width, patches, patches, scale_height, scale_width
|
| 259 |
+
)
|
| 260 |
+
n -= newline_features
|
| 261 |
+
return n
|
| 262 |
+
|
| 263 |
+
def _expand_image_tokens(self, text, image_sizes, height, width, special_token, batch_num_images):
|
| 264 |
+
"""image_mrope=1 multi-image fix: LlavaOV expands each image of a MULTI-image sample as
|
| 265 |
+
`num_image_tokens + 1` (base 576 + one image_newline), bypassing `_get_number_of_features`
|
| 266 |
+
(HF processing_llava_onevision.py). But under image_mrope the model strips image_newline
|
| 267 |
+
(pack_image_features(image_newline=None)) → 576 features/image, so the default +1 over-counts
|
| 268 |
+
by one per image → masked_scatter token/feature mismatch at multi-image eval (e.g. MMMU).
|
| 269 |
+
This override drops the +1 ONLY for image_mrope multi-image; single-image still routes
|
| 270 |
+
through `_get_number_of_features` (AnyRes, newline already stripped there). flag 0 →
|
| 271 |
+
byte-identical to the parent. Verbatim from LlavaOnevisionProcessor._expand_image_tokens
|
| 272 |
+
EXCEPT the gated multi-image count line."""
|
| 273 |
+
image_mrope = video_is_image_mrope(self)
|
| 274 |
+
prompt_strings = []
|
| 275 |
+
max_num_vision_tokens = 0
|
| 276 |
+
for sample in text:
|
| 277 |
+
if special_token in sample:
|
| 278 |
+
num_images = next(batch_num_images) # should consume iterable
|
| 279 |
+
is_multi_image = num_images != 1
|
| 280 |
+
else:
|
| 281 |
+
is_multi_image = False
|
| 282 |
+
while special_token in sample:
|
| 283 |
+
original_size = next(image_sizes) # should consume iterable
|
| 284 |
+
if is_multi_image:
|
| 285 |
+
# ── ONLY deviation: drop the +1 newline under image_mrope (base-only 576).
|
| 286 |
+
num_image_tokens = self.num_image_tokens if image_mrope else self.num_image_tokens + 1
|
| 287 |
+
else:
|
| 288 |
+
if not isinstance(original_size, (list, tuple)):
|
| 289 |
+
# cast to list to avoid numerical precision errors when calculating unpadding
|
| 290 |
+
original_size = original_size.tolist()
|
| 291 |
+
orig_height, orig_width = original_size
|
| 292 |
+
num_image_tokens = self._get_number_of_features(orig_height, orig_width, height, width)
|
| 293 |
+
max_num_vision_tokens = max(max_num_vision_tokens, num_image_tokens)
|
| 294 |
+
if self.vision_feature_select_strategy == "default":
|
| 295 |
+
num_image_tokens -= 1
|
| 296 |
+
sample = sample.replace(special_token, "<placeholder>" * num_image_tokens, 1)
|
| 297 |
+
prompt_strings.append(sample)
|
| 298 |
+
text = [sample.replace("<placeholder>", special_token) for sample in prompt_strings]
|
| 299 |
+
return text, max_num_vision_tokens
|
| 300 |
+
|
| 301 |
+
|
| 302 |
+
__all__ = ["LlavaQwen3_5ViTProcessor"]
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/processor_config.json
ADDED
|
@@ -0,0 +1,209 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"image_processor": {
|
| 3 |
+
"do_convert_rgb": true,
|
| 4 |
+
"do_normalize": true,
|
| 5 |
+
"do_pad": true,
|
| 6 |
+
"do_rescale": true,
|
| 7 |
+
"do_resize": true,
|
| 8 |
+
"image_grid_pinpoints": [
|
| 9 |
+
[
|
| 10 |
+
384,
|
| 11 |
+
384
|
| 12 |
+
],
|
| 13 |
+
[
|
| 14 |
+
384,
|
| 15 |
+
768
|
| 16 |
+
],
|
| 17 |
+
[
|
| 18 |
+
384,
|
| 19 |
+
1152
|
| 20 |
+
],
|
| 21 |
+
[
|
| 22 |
+
384,
|
| 23 |
+
1536
|
| 24 |
+
],
|
| 25 |
+
[
|
| 26 |
+
384,
|
| 27 |
+
1920
|
| 28 |
+
],
|
| 29 |
+
[
|
| 30 |
+
384,
|
| 31 |
+
2304
|
| 32 |
+
],
|
| 33 |
+
[
|
| 34 |
+
768,
|
| 35 |
+
384
|
| 36 |
+
],
|
| 37 |
+
[
|
| 38 |
+
768,
|
| 39 |
+
768
|
| 40 |
+
],
|
| 41 |
+
[
|
| 42 |
+
768,
|
| 43 |
+
1152
|
| 44 |
+
],
|
| 45 |
+
[
|
| 46 |
+
768,
|
| 47 |
+
1536
|
| 48 |
+
],
|
| 49 |
+
[
|
| 50 |
+
768,
|
| 51 |
+
1920
|
| 52 |
+
],
|
| 53 |
+
[
|
| 54 |
+
768,
|
| 55 |
+
2304
|
| 56 |
+
],
|
| 57 |
+
[
|
| 58 |
+
1152,
|
| 59 |
+
384
|
| 60 |
+
],
|
| 61 |
+
[
|
| 62 |
+
1152,
|
| 63 |
+
768
|
| 64 |
+
],
|
| 65 |
+
[
|
| 66 |
+
1152,
|
| 67 |
+
1152
|
| 68 |
+
],
|
| 69 |
+
[
|
| 70 |
+
1152,
|
| 71 |
+
1536
|
| 72 |
+
],
|
| 73 |
+
[
|
| 74 |
+
1152,
|
| 75 |
+
1920
|
| 76 |
+
],
|
| 77 |
+
[
|
| 78 |
+
1152,
|
| 79 |
+
2304
|
| 80 |
+
],
|
| 81 |
+
[
|
| 82 |
+
1536,
|
| 83 |
+
384
|
| 84 |
+
],
|
| 85 |
+
[
|
| 86 |
+
1536,
|
| 87 |
+
768
|
| 88 |
+
],
|
| 89 |
+
[
|
| 90 |
+
1536,
|
| 91 |
+
1152
|
| 92 |
+
],
|
| 93 |
+
[
|
| 94 |
+
1536,
|
| 95 |
+
1536
|
| 96 |
+
],
|
| 97 |
+
[
|
| 98 |
+
1536,
|
| 99 |
+
1920
|
| 100 |
+
],
|
| 101 |
+
[
|
| 102 |
+
1536,
|
| 103 |
+
2304
|
| 104 |
+
],
|
| 105 |
+
[
|
| 106 |
+
1920,
|
| 107 |
+
384
|
| 108 |
+
],
|
| 109 |
+
[
|
| 110 |
+
1920,
|
| 111 |
+
768
|
| 112 |
+
],
|
| 113 |
+
[
|
| 114 |
+
1920,
|
| 115 |
+
1152
|
| 116 |
+
],
|
| 117 |
+
[
|
| 118 |
+
1920,
|
| 119 |
+
1536
|
| 120 |
+
],
|
| 121 |
+
[
|
| 122 |
+
1920,
|
| 123 |
+
1920
|
| 124 |
+
],
|
| 125 |
+
[
|
| 126 |
+
1920,
|
| 127 |
+
2304
|
| 128 |
+
],
|
| 129 |
+
[
|
| 130 |
+
2304,
|
| 131 |
+
384
|
| 132 |
+
],
|
| 133 |
+
[
|
| 134 |
+
2304,
|
| 135 |
+
768
|
| 136 |
+
],
|
| 137 |
+
[
|
| 138 |
+
2304,
|
| 139 |
+
1152
|
| 140 |
+
],
|
| 141 |
+
[
|
| 142 |
+
2304,
|
| 143 |
+
1536
|
| 144 |
+
],
|
| 145 |
+
[
|
| 146 |
+
2304,
|
| 147 |
+
1920
|
| 148 |
+
],
|
| 149 |
+
[
|
| 150 |
+
2304,
|
| 151 |
+
2304
|
| 152 |
+
]
|
| 153 |
+
],
|
| 154 |
+
"image_mean": [
|
| 155 |
+
0.5,
|
| 156 |
+
0.5,
|
| 157 |
+
0.5
|
| 158 |
+
],
|
| 159 |
+
"image_processor_type": "LlavaOnevisionImageProcessor",
|
| 160 |
+
"image_std": [
|
| 161 |
+
0.5,
|
| 162 |
+
0.5,
|
| 163 |
+
0.5
|
| 164 |
+
],
|
| 165 |
+
"resample": 3,
|
| 166 |
+
"rescale_factor": 0.00392156862745098,
|
| 167 |
+
"size": {
|
| 168 |
+
"height": 384,
|
| 169 |
+
"width": 384
|
| 170 |
+
}
|
| 171 |
+
},
|
| 172 |
+
"image_token": "<image>",
|
| 173 |
+
"num_image_tokens": 576,
|
| 174 |
+
"processor_class": "LlavaQwen3_5ViTProcessor",
|
| 175 |
+
"video_mode": "vtc_compression",
|
| 176 |
+
"video_processor": {
|
| 177 |
+
"do_convert_rgb": true,
|
| 178 |
+
"do_normalize": true,
|
| 179 |
+
"do_rescale": true,
|
| 180 |
+
"do_resize": true,
|
| 181 |
+
"do_sample_frames": false,
|
| 182 |
+
"image_mean": [
|
| 183 |
+
0.5,
|
| 184 |
+
0.5,
|
| 185 |
+
0.5
|
| 186 |
+
],
|
| 187 |
+
"image_std": [
|
| 188 |
+
0.5,
|
| 189 |
+
0.5,
|
| 190 |
+
0.5
|
| 191 |
+
],
|
| 192 |
+
"resample": 3,
|
| 193 |
+
"rescale_factor": 0.00392156862745098,
|
| 194 |
+
"return_metadata": false,
|
| 195 |
+
"size": {
|
| 196 |
+
"height": 384,
|
| 197 |
+
"width": 384
|
| 198 |
+
},
|
| 199 |
+
"video_processor_type": "LlavaOnevisionVideoProcessor"
|
| 200 |
+
},
|
| 201 |
+
"video_token": "<video>",
|
| 202 |
+
"vision_aspect_ratio": "anyres_max_9",
|
| 203 |
+
"vision_feature_select_strategy": null,
|
| 204 |
+
"vtc_compression_factor": 2,
|
| 205 |
+
"vtc_temporal_patch_size": 1,
|
| 206 |
+
"auto_map": {
|
| 207 |
+
"AutoProcessor": "processing_qwen3_5vit_qwen3.LlavaQwen3_5ViTProcessor"
|
| 208 |
+
}
|
| 209 |
+
}
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/tokenizer.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9024318c850eaadf26be79389d21b07a7afd8f1af749b89f9109b06c0d12173c
|
| 3 |
+
size 11423018
|
ckpts/S2/8b/qwen3_5_2b/vtc_unique_mkl3_lora/v2-20261025-124726/checkpoint-1533-merged/tokenizer_config.json
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": null,
|
| 5 |
+
"clean_up_tokenization_spaces": false,
|
| 6 |
+
"eos_token": "<|im_end|>",
|
| 7 |
+
"errors": "replace",
|
| 8 |
+
"extra_special_tokens": [
|
| 9 |
+
"<image>",
|
| 10 |
+
"<video>"
|
| 11 |
+
],
|
| 12 |
+
"is_local": true,
|
| 13 |
+
"model_max_length": 131072,
|
| 14 |
+
"pad_token": "<|endoftext|>",
|
| 15 |
+
"processor_class": "LlavaQwen3_5ViTProcessor",
|
| 16 |
+
"split_special_tokens": false,
|
| 17 |
+
"tokenizer_class": "Qwen2Tokenizer",
|
| 18 |
+
"unk_token": null
|
| 19 |
+
}
|