Add six S2 vision backbone baselines

.gitattributes

@@ -94,3 +94,9 @@ ckpts/S2/4b/qwen3_5_2b/v13_6/v0-20260820-192228/checkpoint-1400/tokenizer.json f
 ckpts/S2/4b/qwen3_5_2b/v14_1_mkl_layer3/v0-20261012-043050/checkpoint-1530/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 ckpts/S2/4b/qwen3_5_2b/baseline_tmerge_64f_20pct/v0-20261012-171609/checkpoint-1530/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 ckpts/S2/4b/siglip2_large/siglip2_large/v0-20261024-091108/checkpoint-1530/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+ckpts/S2/4b/pe_lang_large/pe_lang_large/v0-20261024-152414/checkpoint-2182/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+ckpts/S2/4b/radio_v3_large/radio_v3_large/v0-20261024-142322/checkpoint-1530/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+ckpts/S2/4b/siglip2_so400m/siglip2_so400m/v0-20261024-102319/checkpoint-2182/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+ckpts/S2/4b/siglip_hd/siglip_hd/v0-20261024-104810/checkpoint-1530/tokenizer.json filter=lfs diff=lfs merge=lfs -text

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/args.json

@@ -0,0 +1,393 @@
+{
+  "output_dir": "/home/wangjunjie09/VisionEncoder/V3_exps/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505",
+  "per_device_train_batch_size": 1,
+  "num_train_epochs": 1.0,
+  "max_steps": -1,
+  "learning_rate": 1e-05,
+  "lr_scheduler_type": "cosine",
+  "lr_scheduler_kwargs": null,
+  "warmup_steps": 0,
+  "optim": "adamw_torch_fused",
+  "optim_args": null,
+  "weight_decay": 0.1,
+  "adam_beta1": 0.9,
+  "adam_beta2": 0.95,
+  "adam_epsilon": 1e-08,
+  "optim_target_modules": null,
+  "gradient_accumulation_steps": 8,
+  "average_tokens_across_devices": true,
+  "max_grad_norm": 1.0,
+  "label_smoothing_factor": 0.0,
+  "bf16": true,
+  "fp16": false,
+  "bf16_full_eval": false,
+  "fp16_full_eval": false,
+  "tf32": null,
+  "gradient_checkpointing": true,
+  "gradient_checkpointing_kwargs": "{\"use_reentrant\": false}",
+  "torch_compile": false,
+  "torch_compile_backend": null,
+  "torch_compile_mode": null,
+  "use_liger_kernel": false,
+  "liger_kernel_config": null,
+  "use_cache": false,
+  "neftune_noise_alpha": null,
+  "torch_empty_cache_steps": null,
+  "auto_find_batch_size": false,
+  "logging_strategy": "steps",
+  "logging_steps": 5,
+  "logging_first_step": true,
+  "log_on_each_node": true,
+  "logging_nan_inf_filter": true,
+  "include_num_input_tokens_seen": false,
+  "log_level": "passive",
+  "log_level_replica": "warning",
+  "disable_tqdm": null,
+  "report_to": [
+    "none"
+  ],
+  "run_name": "/home/wangjunjie09/exps/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505",
+  "project": "huggingface",
+  "trackio_space_id": "trackio",
+  "eval_strategy": "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": null,
+  "include_for_metrics": [],
+  "batch_eval_metrics": false,
+  "save_only_model": false,
+  "save_strategy": "steps",
+  "save_steps": 100.0,
+  "save_on_each_node": false,
+  "save_total_limit": 1,
+  "enable_jit_checkpoint": false,
+  "push_to_hub": false,
+  "hub_token": null,
+  "hub_private_repo": null,
+  "hub_model_id": null,
+  "hub_strategy": "every_save",
+  "hub_always_push": false,
+  "hub_revision": null,
+  "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": "{\"dispatch_batches\": false}",
+  "parallelism_config": null,
+  "dataloader_drop_last": false,
+  "dataloader_num_workers": 8,
+  "dataloader_pin_memory": true,
+  "dataloader_persistent_workers": true,
+  "dataloader_prefetch_factor": 4,
+  "remove_unused_columns": true,
+  "label_names": null,
+  "train_sampling_strategy": "random",
+  "length_column_name": "length",
+  "ddp_find_unused_parameters": null,
+  "ddp_bucket_cap_mb": null,
+  "ddp_broadcast_buffers": null,
+  "ddp_backend": null,
+  "ddp_timeout": 7200,
+  "fsdp": [],
+  "fsdp_config": null,
+  "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": null,
+  "skip_memory_metrics": true,
+  "do_train": false,
+  "do_eval": false,
+  "do_predict": false,
+  "resume_from_checkpoint": null,
+  "warmup_ratio": 0.05,
+  "logging_dir": "/home/wangjunjie09/exps/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/runs",
+  "local_rank": 0,
+  "sortish_sampler": false,
+  "predict_with_generate": false,
+  "generation_max_length": null,
+  "generation_num_beams": null,
+  "generation_config": null,
+  "tuner_backend": "peft",
+  "llm_checkpointing": 16,
+  "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": null,
+  "aligner_lr": null,
+  "vit_lr": 1e-06,
+  "use_logits_to_keep": true,
+  "ds3_gather_for_generation": true,
+  "resume_only_model": false,
+  "optimizer": null,
+  "loss_type": null,
+  "eval_metric": null,
+  "callbacks": [],
+  "early_stop_interval": null,
+  "eval_use_evalscope": false,
+  "eval_dataset": [],
+  "eval_dataset_args": null,
+  "eval_limit": null,
+  "eval_generation_config": null,
+  "extra_eval_args": null,
+  "tuner_type": "full",
+  "use_galore": false,
+  "galore_target_modules": null,
+  "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,
+  "use_ray": false,
+  "ray_exp_name": null,
+  "device_groups": null,
+  "model": "/home/wangjunjie09/VisionEncoder/V3_exps/S1/4b/aimv2_large/aimv2_large/v0-20261024-150052/checkpoint-2181",
+  "model_type": "llava_external_vit_qwen3",
+  "model_revision": null,
+  "task_type": "causal_lm",
+  "torch_dtype": "bfloat16",
+  "attn_impl": "flash_attn",
+  "experts_impl": null,
+  "new_special_tokens": [],
+  "num_labels": null,
+  "problem_type": null,
+  "rope_scaling": null,
+  "device_map": null,
+  "max_memory": {},
+  "max_model_len": null,
+  "local_repo_path": null,
+  "init_strategy": null,
+  "template": "llava_external_vit_qwen3",
+  "system": null,
+  "max_length": 16384,
+  "truncation_strategy": "delete",
+  "max_pixels": null,
+  "agent_template": null,
+  "norm_bbox": null,
+  "use_chat_template": true,
+  "padding_side": "right",
+  "padding_free": true,
+  "loss_scale": "default",
+  "sequence_parallel_size": 1,
+  "template_backend": "swift",
+  "response_prefix": null,
+  "enable_thinking": null,
+  "add_non_thinking_prefix": true,
+  "dataset": [],
+  "val_dataset": [],
+  "cached_dataset": [
+    "/share/m2v_intern/wangjunjie09/datasets/prepared/s2/img729_vid196/image_20pct/train",
+    "/share/m2v_intern/wangjunjie09/datasets/prepared/s2/img729_vid196/video_20pct/v2/train"
+  ],
+  "cached_val_dataset": [],
+  "split_dataset_ratio": 0.0,
+  "dataset_num_proc": 16,
+  "load_from_cache_file": false,
+  "dataset_shuffle": true,
+  "val_dataset_shuffle": false,
+  "streaming": false,
+  "interleave_prob": null,
+  "stopping_strategy": "first_exhausted",
+  "shuffle_buffer_size": 1000,
+  "download_mode": "reuse_dataset_if_exists",
+  "columns": {},
+  "strict": false,
+  "model_name": null,
+  "model_author": null,
+  "custom_dataset_info": [],
+  "quant_method": null,
+  "quant_bits": null,
+  "hqq_axis": null,
+  "bnb_4bit_compute_dtype": "bfloat16",
+  "bnb_4bit_quant_type": "nf4",
+  "bnb_4bit_use_double_quant": true,
+  "bnb_4bit_quant_storage": null,
+  "max_new_tokens": 64,
+  "temperature": 0.0,
+  "top_k": null,
+  "top_p": null,
+  "repetition_penalty": null,
+  "num_beams": 1,
+  "stream": false,
+  "stop_words": [],
+  "logprobs": false,
+  "top_logprobs": null,
+  "structured_outputs_regex": null,
+  "train_type": null,
+  "adapters": [],
+  "external_plugins": [
+    "video_mllm/training/model_plugin_external_vit.py",
+    "video_mllm/data/dataset_plugin.py",
+    "/home/wangjunjie09/VisionEncoder/video_mllm/training/log_spacing_plugin.py"
+  ],
+  "custom_register_path": [
+    "/home/wangjunjie09/VisionEncoder/video_mllm/training/log_spacing_plugin.py"
+  ],
+  "model_kwargs": {
+    "vision_source_id": "aimv2_large_patch14_336_distilled",
+    "num_frames": 32,
+    "eval_frames": 32,
+    "s2_use_video_frame_cache": true,
+    "s2_video_frame_cache": "/share/m2v_intern_v3/wangjunjie09/datasets/video_cache/s2/frames/jpeg/384/32"
+  },
+  "video_cache": {},
+  "image_teacher_models": {},
+  "load_args": false,
+  "load_data_args": false,
+  "packing": true,
+  "packing_length": 16384,
+  "packing_num_proc": 1,
+  "lazy_tokenize": false,
+  "use_hf": true,
+  "ignore_args_error": false,
+  "use_swift_lora": false,
+  "freeze_parameters": [],
+  "freeze_parameters_regex": null,
+  "freeze_parameters_ratio": 0.0,
+  "trainable_parameters": [
+    "model.multi_modal_projector"
+  ],
+  "trainable_parameters_regex": null,
+  "freeze_llm": false,
+  "freeze_vit": false,
+  "freeze_aligner": false,
+  "target_modules": [
+    "all-linear"
+  ],
+  "target_regex": null,
+  "target_parameters": null,
+  "modules_to_save": [],
+  "lora_rank": 8,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "lora_bias": "none",
+  "lora_dtype": null,
+  "lorap_lr_ratio": null,
+  "use_rslora": false,
+  "use_dora": false,
+  "lora_ga_batch_size": 2,
+  "lora_ga_iters": 2,
+  "lora_ga_max_length": 1024,
+  "lora_ga_direction": "ArB2r",
+  "lora_ga_scale": "stable",
+  "lora_ga_stable_gamma": 16,
+  "init_weights": true,
+  "fourier_n_frequency": 2000,
+  "fourier_scaling": 300.0,
+  "boft_block_size": 4,
+  "boft_block_num": 0,
+  "boft_n_butterfly_factor": 1,
+  "boft_dropout": 0.0,
+  "vera_rank": 256,
+  "vera_projection_prng_key": 0,
+  "vera_dropout": 0.0,
+  "vera_d_initial": 0.1,
+  "adapter_act": "gelu",
+  "adapter_length": 128,
+  "adalora_target_r": 8,
+  "adalora_init_r": 12,
+  "adalora_tinit": 0,
+  "adalora_tfinal": 0,
+  "adalora_deltaT": 1,
+  "adalora_beta1": 0.85,
+  "adalora_beta2": 0.85,
+  "adalora_orth_reg_weight": 0.5,
+  "llamapro_num_new_blocks": 4,
+  "llamapro_num_groups": null,
+  "reft_layer_key": null,
+  "reft_layers": null,
+  "reft_rank": 4,
+  "reft_intervention_type": "LoreftIntervention",
+  "reft_args": null,
+  "swanlab_token": null,
+  "swanlab_project": "ms-swift",
+  "swanlab_workspace": null,
+  "swanlab_exp_name": null,
+  "swanlab_notification_method": null,
+  "swanlab_webhook_url": null,
+  "swanlab_secret": null,
+  "swanlab_sender_email": null,
+  "swanlab_receiver_email": null,
+  "swanlab_smtp_server": null,
+  "swanlab_smtp_port": null,
+  "swanlab_email_language": "zh",
+  "swanlab_mode": "cloud",
+  "add_version": true,
+  "create_checkpoint_symlink": false,
+  "v12_resolution": 384,
+  "zero_hpz_partition_size": null,
+  "deepspeed_autotp_size": null,
+  "swift_version": "4.1.0.dev0",
+  "ckpt_dir": "/home/wangjunjie09/VisionEncoder/V3_exps/S1/4b/aimv2_large/aimv2_large/v0-20261024-150052/checkpoint-2181",
+  "rank": 0,
+  "global_world_size": 8,
+  "local_world_size": 8,
+  "model_suffix": "checkpoint-2181",
+  "model_info": "ModelInfo(model_type='llava_external_vit_qwen3', model_dir='/home/wangjunjie09/VisionEncoder/exps/S1/4b/aimv2_large/aimv2_large/v0-20261024-150052/checkpoint-2181', torch_dtype=torch.bfloat16, max_model_len=262144, 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)",
+  "model_meta": "ModelMeta(model_type='llava_external_vit_qwen3', model_groups=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen3-4B-Instruct-2507', hf_model_id='Qwen/Qwen3-4B-Instruct-2507', model_path=None, ms_revision=None, hf_revision=None)], template=None, ignore_patterns=None, requires=None, tags=[])], loader=<class 'model_plugin_external_vit.ExternalViTLoader'>, template='llava_external_vit_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=['LlavaExternalViTForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=[], tags=['vision', 'video'])",
+  "model_dir": "/home/wangjunjie09/VisionEncoder/V3_exps/S1/4b/aimv2_large/aimv2_large/v0-20261024-150052/checkpoint-2181",
+  "template_meta": "QwenTemplateMeta(template_type='llava_external_vit_qwen3', 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 'model_plugin_external_vit.ExternalViTTemplate'>, 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='', history_thinking_prefix='')",
+  "_val_dataset_exists": false,
+  "hub": "<class 'swift.hub.hub.HFHub'>",
+  "evaluation_strategy": "steps",
+  "training_args": "Seq2SeqTrainingArguments(output_dir='/home/wangjunjie09/exps/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505', per_device_train_batch_size=1, num_train_epochs=1.0, max_steps=-1, learning_rate=1e-05, 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/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505', 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/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/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=16, 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=None, 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='full', 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)"
+}

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/chat_template.jinja

@@ -0,0 +1,61 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/chat_template.json

@@ -0,0 +1,3 @@
+{
+  "chat_template": "{% for message in messages %}{{'<|im_start|>' + message['role'] + ' '}}{# Render all images first #}{% for content in message['content'] | selectattr('type', 'equalto', 'image') %}{{ '<image>' }}{% endfor %}{# Render all video then #}{% for content in message['content'] | selectattr('type', 'equalto', 'video') %}{{ '<video>' }}{% endfor %}{# Render all text next #}{% if message['role'] != 'assistant' %}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{{ '\n' + content['text'] }}{% endfor %}{% else %}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{% generation %}{{ '\n' + content['text'] }}{% endgeneration %}{% endfor %}{% endif %}{{'<|im_end|>'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}"
+}

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/config.json

@@ -0,0 +1,287 @@
+{
+  "architectures": [
+    "LlavaExternalViTForConditionalGeneration"
+  ],
+  "bos_token_id": null,
+  "dtype": "bfloat16",
+  "eos_token_id": 151645,
+  "eval_num_frames": 32,
+  "hidden_size": 2560,
+  "image_grid_pinpoints": [
+    [
+      384,
+      384
+    ],
+    [
+      384,
+      768
+    ],
+    [
+      384,
+      1152
+    ],
+    [
+      384,
+      1536
+    ],
+    [
+      384,
+      1920
+    ],
+    [
+      384,
+      2304
+    ],
+    [
+      768,
+      384
+    ],
+    [
+      768,
+      768
+    ],
+    [
+      768,
+      1152
+    ],
+    [
+      768,
+      1536
+    ],
+    [
+      768,
+      1920
+    ],
+    [
+      768,
+      2304
+    ],
+    [
+      1152,
+      384
+    ],
+    [
+      1152,
+      768
+    ],
+    [
+      1152,
+      1152
+    ],
+    [
+      1152,
+      1536
+    ],
+    [
+      1152,
+      1920
+    ],
+    [
+      1152,
+      2304
+    ],
+    [
+      1536,
+      384
+    ],
+    [
+      1536,
+      768
+    ],
+    [
+      1536,
+      1152
+    ],
+    [
+      1536,
+      1536
+    ],
+    [
+      1536,
+      1920
+    ],
+    [
+      1536,
+      2304
+    ],
+    [
+      1920,
+      384
+    ],
+    [
+      1920,
+      768
+    ],
+    [
+      1920,
+      1152
+    ],
+    [
+      1920,
+      1536
+    ],
+    [
+      1920,
+      1920
+    ],
+    [
+      1920,
+      2304
+    ],
+    [
+      2304,
+      384
+    ],
+    [
+      2304,
+      768
+    ],
+    [
+      2304,
+      1152
+    ],
+    [
+      2304,
+      1536
+    ],
+    [
+      2304,
+      1920
+    ],
+    [
+      2304,
+      2304
+    ]
+  ],
+  "image_token_index": 151669,
+  "input_resolution": 384,
+  "model_type": "llava_external_vit_qwen3",
+  "multimodal_projector_bias": true,
+  "pad_token_id": 151643,
+  "projector_hidden_act": "gelu",
+  "text_config": {
+    "_name_or_path": "/share/m2v_intern_v3/wangjunjie09/model_cache/huggingface/Qwen/Qwen3-4B-Instruct-2507",
+    "architectures": [
+      "Qwen3ForCausalLM"
+    ],
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "bos_token_id": 151643,
+    "dtype": "bfloat16",
+    "eos_token_id": 151645,
+    "head_dim": 128,
+    "hidden_act": "silu",
+    "hidden_size": 2560,
+    "initializer_range": 0.02,
+    "intermediate_size": 9728,
+    "layer_types": [
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention"
+    ],
+    "max_position_embeddings": 262144,
+    "max_window_layers": 36,
+    "model_type": "qwen3",
+    "num_attention_heads": 32,
+    "num_hidden_layers": 36,
+    "num_key_value_heads": 8,
+    "pad_token_id": 151643,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+      "rope_theta": 5000000,
+      "rope_type": "default"
+    },
+    "sliding_window": null,
+    "tie_word_embeddings": true,
+    "use_cache": false,
+    "use_sliding_window": false,
+    "vocab_size": 151936
+  },
+  "tie_word_embeddings": true,
+  "train_num_frames": 32,
+  "transformers_version": "5.5.4",
+  "use_cache": false,
+  "video_token_index": 151670,
+  "vision_aspect_ratio": "anyres_max_9",
+  "vision_config": {
+    "dtype": "bfloat16",
+    "external_vit_type": "aimv2_large",
+    "hidden_size": 1024,
+    "image_size": 384,
+    "model_type": "external_vit",
+    "patch_size": 14,
+    "source_image_size": 336,
+    "upstream_config": {
+      "architectures": [
+        "Aimv2VisionModel"
+      ],
+      "attention_dropout": 0.0,
+      "auto_map": {
+        "AutoConfig": "configuration_aimv2.AIMv2Config",
+        "AutoModel": "modeling_aimv2.AIMv2Model",
+        "FlaxAutoModel": "modeling_flax_aimv2.FlaxAIMv2Model"
+      },
+      "hidden_act": "silu",
+      "hidden_size": 1024,
+      "image_size": 336,
+      "initializer_range": 0.02,
+      "intermediate_size": 2816,
+      "is_native": false,
+      "mlp_bias": false,
+      "model_type": "aimv2_vision_model",
+      "num_attention_heads": 8,
+      "num_channels": 3,
+      "num_hidden_layers": 24,
+      "patch_size": 14,
+      "projection_dropout": 0.0,
+      "qkv_bias": false,
+      "rms_norm_eps": 1e-05,
+      "torch_dtype": "float32",
+      "transformers_version": "4.54.0.dev0",
+      "use_bias": false,
+      "use_head": false
+    }
+  },
+  "vision_family": "aimv2_large",
+  "vision_feature_layer": -1,
+  "vision_feature_select_strategy": "full",
+  "auto_map": {
+    "AutoConfig": "modeling_external_vit_qwen3.LlavaExternalViTConfig",
+    "AutoModel": "modeling_external_vit_qwen3.LlavaExternalViTForConditionalGeneration",
+    "AutoModelForCausalLM": "modeling_external_vit_qwen3.LlavaExternalViTForConditionalGeneration",
+    "AutoModelForImageTextToText": "modeling_external_vit_qwen3.LlavaExternalViTForConditionalGeneration"
+  }
+}

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/external_vit/__init__.py

@@ -0,0 +1,5 @@
+"""Self-contained external vision backbones used by the Qwen3 integration."""
+
+from .backbones import ExternalViTBackbone, load_external_vit_source
+
+__all__ = ["ExternalViTBackbone", "load_external_vit_source"]

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/external_vit/backbones.py

@@ -0,0 +1,121 @@
+"""Construction and source loading for supported external vision towers."""
+
+import json
+import math
+from pathlib import Path
+from typing import Any
+
+import torch
+from safetensors.torch import load_file
+from torch import nn
+from torch.nn import functional as F
+from transformers.models.aimv2 import Aimv2VisionConfig, Aimv2VisionModel
+
+
+PE_LANG = "pe_lang_large"
+C_RADIO = "radio_v3_large"
+AIMV2 = "aimv2_large"
+
+
+def source_metadata(source: str | Path) -> dict[str, Any]:
+    """Read the architecture metadata needed to rebuild a saved tower."""
+    source = Path(source)
+    config_path = source / "config.json"
+    if not config_path.is_file():
+        return {
+            "external_vit_type": PE_LANG,
+            "hidden_size": 1024,
+            "patch_size": 14,
+            "source_image_size": 448,
+            "upstream_config": {},
+        }
+
+    upstream_config = json.loads(config_path.read_text())
+    if upstream_config.get("version") == "c-radio_v3-l":
+        return {
+            "external_vit_type": C_RADIO,
+            "hidden_size": 1024,
+            "patch_size": 16,
+            "source_image_size": 512,
+            "upstream_config": upstream_config,
+        }
+    return {
+        "external_vit_type": AIMV2,
+        "hidden_size": upstream_config["hidden_size"],
+        "patch_size": upstream_config["patch_size"],
+        "source_image_size": upstream_config["image_size"],
+        "upstream_config": upstream_config,
+    }
+
+
+class ExternalViTBackbone(nn.Module):
+    """Expose each upstream tower's canonical final spatial readout."""
+
+    def __init__(self, config):
+        super().__init__()
+        self.external_vit_type = config.external_vit_type
+        upstream = dict(config.upstream_config)
+        if self.external_vit_type == PE_LANG:
+            from .perception_models.core.vision_encoder.pe import VisionTransformer
+
+            self.vision = VisionTransformer.from_config("PE-Lang-L14-448-Tiling")
+            # RoPE owns unregistered tensors, so HF meta loading cannot materialize them.
+            with torch.device("cpu"):
+                self.vision.rope.init_tensors()
+        elif self.external_vit_type == C_RADIO:
+            from .radio import RADIOConfig, RADIOModel
+
+            # Honor the dtype selected by the enclosing model's construction context.
+            upstream["args"] = {
+                **upstream["args"],
+                "dtype": str(torch.get_default_dtype()).split(".")[-1],
+            }
+            self.vision = RADIOModel(RADIOConfig(**upstream))
+        elif self.external_vit_type == AIMV2:
+            upstream["image_size"] = config.image_size
+            self.vision = Aimv2VisionModel(Aimv2VisionConfig(**upstream))
+        else:
+            raise ValueError(f"Unsupported external ViT type: {self.external_vit_type}")
+
+    def forward(self, pixels: torch.Tensor) -> torch.Tensor:
+        if self.external_vit_type == PE_LANG:
+            return self.vision.forward_features(pixels, strip_cls_token=True)
+        if self.external_vit_type == C_RADIO:
+            self.vision.input_conditioner.dtype = next(self.vision.model.parameters()).dtype
+            return self.vision(pixels).features
+        return self.vision(pixel_values=pixels, return_dict=True).last_hidden_state
+
+
+def _resize_aim_position_embedding(
+    position_embedding: torch.Tensor, target_grid: int
+) -> torch.Tensor:
+    source_grid = math.isqrt(position_embedding.shape[0])
+    position_embedding = position_embedding.reshape(
+        1, source_grid, source_grid, -1
+    ).permute(0, 3, 1, 2)
+    position_embedding = F.interpolate(
+        position_embedding.float(),
+        size=(target_grid, target_grid),
+        mode="bicubic",
+        align_corners=False,
+    ).to(position_embedding.dtype)
+    return position_embedding.permute(0, 2, 3, 1).reshape(target_grid**2, -1)
+
+
+def load_external_vit_source(
+    backbone: ExternalViTBackbone, source: str | Path, target_grid: int
+) -> None:
+    """Load upstream weights; PE interpolates its native position grid at runtime."""
+    source = Path(source)
+    if backbone.external_vit_type == PE_LANG:
+        state = torch.load(
+            source / "PE-Lang-L14-448-Tiling.pt",
+            map_location="cpu",
+            weights_only=True,
+        )
+    else:
+        state = load_file(source / "model.safetensors")
+        if backbone.external_vit_type == AIMV2:
+            key = "embeddings.position_embedding.weight"
+            state[key] = _resize_aim_position_embedding(state[key], target_grid)
+    backbone.vision.load_state_dict(state)

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/generation_config.json

@@ -0,0 +1,12 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 151643,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "transformers_version": "5.5.4",
+  "use_cache": true
+}

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/latest

@@ -0,0 +1 @@
+global_step2182

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/model-00001-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f04e692bf9c2d59c02edac2ff7c760bff30413d96d5c4cddc63b611897cefae7
+size 4987507408

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/model-00002-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2405ccf9d2fd56866b37a8821d704423a4e4da85804638ab41500dea57200b7
+size 4473156208

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/model.safetensors.index.json

@@ -0,0 +1,633 @@
+{
+  "metadata": {
+    "total_parameters": 4341332992,
+    "total_size": 9460578304
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00001-of-00002.safetensors",
+    "model.image_newline": "model-00001-of-00002.safetensors",
+    "model.language_model.embed_tokens.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.17.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.30.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.norm.weight": "model-00002-of-00002.safetensors",
+    "model.multi_modal_projector.linear_1.bias": "model-00002-of-00002.safetensors",
+    "model.multi_modal_projector.linear_1.weight": "model-00002-of-00002.safetensors",
+    "model.multi_modal_projector.linear_2.bias": "model-00002-of-00002.safetensors",
+    "model.multi_modal_projector.linear_2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.embeddings.patch_embed.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.embeddings.patch_embed.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.embeddings.position_embedding.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.embeddings.rms_norm.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.0.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.0.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.0.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.0.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.0.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.0.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.0.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.0.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.0.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.1.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.1.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.1.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.1.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.1.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.1.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.1.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.1.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.1.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.10.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.10.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.10.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.10.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.10.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.10.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.10.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.10.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.10.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.11.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.11.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.11.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.11.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.11.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.11.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.11.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.11.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.11.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.12.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.12.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.12.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.12.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.12.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.12.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.12.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.12.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.12.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.13.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.13.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.13.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.13.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.13.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.13.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.13.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.13.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.13.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.14.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.14.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.14.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.14.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.14.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.14.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.14.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.14.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.14.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.15.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.15.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.15.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.15.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.15.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.15.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.15.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.15.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.15.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.16.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.16.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.16.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.16.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.16.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.16.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.16.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.16.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.16.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.17.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.17.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.17.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.17.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.17.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.17.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.17.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.17.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.17.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.18.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.18.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.18.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.18.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.18.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.18.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.18.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.18.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.18.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.19.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.19.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.19.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.19.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.19.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.19.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.19.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.19.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.19.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.2.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.2.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.2.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.2.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.2.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.2.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.2.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.2.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.2.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.20.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.20.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.20.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.20.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.20.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.20.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.20.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.20.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.20.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.21.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.21.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.21.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.21.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.21.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.21.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.21.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.21.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.21.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.22.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.22.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.22.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.22.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.22.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.22.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.22.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.22.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.22.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.23.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.23.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.23.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.23.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.23.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.23.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.23.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.23.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.23.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.3.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.3.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.3.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.3.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.3.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.3.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.3.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.3.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.3.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.4.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.4.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.4.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.4.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.4.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.4.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.4.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.4.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.4.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.5.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.5.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.5.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.5.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.5.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.5.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.5.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.5.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.5.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.6.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.6.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.6.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.6.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.6.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.6.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.6.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.6.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.6.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.7.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.7.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.7.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.7.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.7.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.7.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.7.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.7.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.7.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.8.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.8.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.8.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.8.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.8.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.8.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.8.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.8.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.8.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.9.attention.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.9.attention.out_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.9.attention.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.9.attention.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.9.ffn.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.9.ffn.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.9.ffn.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.9.rms_norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.encoder.layers.9.rms_norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.vision.rms_norm.weight": "model-00002-of-00002.safetensors"
+  }
+}

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/modeling_external_vit_qwen3.py

@@ -0,0 +1,208 @@
+"""External vision towers with LLaVA-OneVision packing and Qwen3."""
+
+import math
+from types import SimpleNamespace
+
+import torch
+from torch import nn
+from torch.nn import functional as F
+from transformers import AutoConfig, AutoModel
+from transformers.configuration_utils import PretrainedConfig
+from transformers.modeling_outputs import BaseModelOutputWithPooling
+from transformers.models.llava_onevision.configuration_llava_onevision import (
+    LlavaOnevisionConfig,
+)
+from transformers.models.llava_onevision.modeling_llava_onevision import (
+    LlavaOnevisionForConditionalGeneration,
+    LlavaOnevisionModel,
+    LlavaOnevisionMultiModalProjector,
+    LlavaOnevisionPreTrainedModel,
+    image_size_to_num_patches,
+)
+
+from external_vit import ExternalViTBackbone
+from external_vit.backbones import C_RADIO
+
+
+class ExternalViTConfig(PretrainedConfig):
+    model_type = "external_vit"
+
+    def __init__(
+        self,
+        external_vit_type="pe_lang_large",
+        hidden_size=1024,
+        patch_size=14,
+        image_size=384,
+        source_image_size=448,
+        upstream_config=None,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+        self.external_vit_type = external_vit_type
+        self.hidden_size = hidden_size
+        self.patch_size = patch_size
+        self.image_size = image_size
+        self.source_image_size = source_image_size
+        self.upstream_config = upstream_config or {}
+
+
+AutoConfig.register(ExternalViTConfig.model_type, ExternalViTConfig, exist_ok=True)
+
+
+class LlavaExternalViTConfig(LlavaOnevisionConfig):
+    model_type = "llava_external_vit_qwen3"
+
+    def __init__(
+        self,
+        vision_config=None,
+        vision_family=None,
+        input_resolution=384,
+        train_num_frames=32,
+        eval_num_frames=32,
+        vision_aspect_ratio="anyres_max_9",
+        image_grid_pinpoints=None,
+        **kwargs,
+    ):
+        if vision_config is None:
+            vision_config = ExternalViTConfig(image_size=input_resolution)
+        if isinstance(vision_config, dict):
+            vision_config = ExternalViTConfig(**vision_config)
+        if image_grid_pinpoints is None:
+            image_grid_pinpoints = [
+                [input_resolution * height, input_resolution * width]
+                for height in range(1, 7)
+                for width in range(1, 7)
+            ]
+        super().__init__(
+            vision_config=vision_config,
+            vision_aspect_ratio=vision_aspect_ratio,
+            image_grid_pinpoints=image_grid_pinpoints,
+            **kwargs,
+        )
+        self.input_resolution = input_resolution
+        self.vision_family = vision_family or vision_config.external_vit_type
+        self.train_num_frames = train_num_frames
+        self.eval_num_frames = eval_num_frames
+
+
+class ExternalViTTower(nn.Module):
+    def __init__(self, config: LlavaExternalViTConfig):
+        super().__init__()
+        self.vision = ExternalViTBackbone(config.vision_config)
+
+    def forward(self, pixels: torch.Tensor, *, video: bool = False) -> torch.Tensor:
+        leading = pixels.shape[:-3]
+        pixels = pixels.flatten(0, -4) if video else pixels
+        if video and self.vision.external_vit_type == C_RADIO:
+            with self.vision.vision.radio_model.cpe_video_mode(t=leading[-1]):
+                features = self.vision(pixels)
+        else:
+            features = self.vision(pixels)
+        return features.reshape(*leading, *features.shape[-2:])
+
+
+class LlavaExternalViTModel(LlavaOnevisionModel):
+    config_class = LlavaExternalViTConfig
+
+    def __init__(self, config: LlavaExternalViTConfig):
+        LlavaOnevisionPreTrainedModel.__init__(self, config)
+        self.vision_tower = ExternalViTTower(config)
+        self.multi_modal_projector = LlavaOnevisionMultiModalProjector(config)
+        self.image_newline = nn.Parameter(
+            torch.randn(config.text_config.hidden_size)
+            / math.sqrt(config.text_config.hidden_size)
+        )
+        self.vocab_size = config.text_config.vocab_size
+        self.language_model = AutoModel.from_config(config.text_config)
+        self.post_init()
+
+    def get_image_features(self, pixel_values, image_sizes=None, **kwargs):
+        if self.config.vision_aspect_ratio == "square":
+            pixels = pixel_values[:, 0] if pixel_values.ndim == 5 else pixel_values
+            features = self.multi_modal_projector(self.vision_tower(pixels))
+            newline = self.image_newline[None, None].expand(features.shape[0], 1, -1)
+            features = torch.cat((features, newline.to(features)), dim=1)
+            return BaseModelOutputWithPooling(
+                pooler_output=list(features.unbind(0))
+            )
+
+        tile_counts = [
+            image_size_to_num_patches(
+                size, self.config.image_grid_pinpoints, self.config.input_resolution
+            )
+            for size in image_sizes
+        ]
+        if pixel_values.ndim == 5:
+            pixel_values = torch.cat(
+                [pixels[:count] for pixels, count in zip(pixel_values, tile_counts)]
+            )
+        features = self.multi_modal_projector(self.vision_tower(pixel_values))
+        features, _ = self.pack_image_features(
+            features.split(tile_counts),
+            image_sizes,
+            image_newline=self.image_newline,
+            vision_aspect_ratio=self.config.vision_aspect_ratio,
+        )
+        return BaseModelOutputWithPooling(pooler_output=features)
+
+    def pack_image_features(
+        self,
+        image_features,
+        image_sizes,
+        image_newline=None,
+        vision_aspect_ratio="anyres_max_9",
+    ):
+        geometry = SimpleNamespace(
+            config=SimpleNamespace(
+                vision_config=SimpleNamespace(
+                    image_size=self.config.input_resolution,
+                    patch_size=self.config.vision_config.patch_size,
+                ),
+                image_grid_pinpoints=self.config.image_grid_pinpoints,
+            )
+        )
+        return LlavaOnevisionModel.pack_image_features(
+            geometry,
+            image_features,
+            image_sizes,
+            image_newline=image_newline,
+            vision_aspect_ratio=vision_aspect_ratio,
+        )
+
+    def get_video_features(self, pixel_values, **kwargs):
+        features = self.multi_modal_projector(
+            self.vision_tower(pixel_values, video=True)
+        )
+        batch, frames, patches, width = features.shape
+        side = math.isqrt(patches)
+        features = features.reshape(
+            batch * frames, side, side, width
+        ).permute(0, 3, 1, 2)
+        features = F.interpolate(
+            features,
+            size=(math.ceil(side / 2), math.ceil(side / 2)),
+            mode="bilinear",
+        )
+        features = features.permute(0, 2, 3, 1).reshape(batch, -1, width)
+        return BaseModelOutputWithPooling(pooler_output=features)
+
+
+class LlavaExternalViTForConditionalGeneration(
+    LlavaOnevisionForConditionalGeneration
+):
+    config_class = LlavaExternalViTConfig
+
+    def __init__(self, config: LlavaExternalViTConfig):
+        LlavaOnevisionPreTrainedModel.__init__(self, config)
+        self.model = LlavaExternalViTModel(config)
+        self.lm_head = nn.Linear(
+            config.text_config.hidden_size,
+            config.text_config.vocab_size,
+            bias=False,
+        )
+        self.post_init()
+
+
+AutoConfig.register(
+    LlavaExternalViTConfig.model_type, LlavaExternalViTConfig, exist_ok=True
+)

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/processor_config.json

@@ -0,0 +1,213 @@
+{
+  "image_processor": {
+    "crop_size": {
+      "height": 384,
+      "width": 384
+    },
+    "do_center_crop": true,
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_pad": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_grid_pinpoints": [
+      [
+        384,
+        384
+      ],
+      [
+        384,
+        768
+      ],
+      [
+        384,
+        1152
+      ],
+      [
+        384,
+        1536
+      ],
+      [
+        384,
+        1920
+      ],
+      [
+        384,
+        2304
+      ],
+      [
+        768,
+        384
+      ],
+      [
+        768,
+        768
+      ],
+      [
+        768,
+        1152
+      ],
+      [
+        768,
+        1536
+      ],
+      [
+        768,
+        1920
+      ],
+      [
+        768,
+        2304
+      ],
+      [
+        1152,
+        384
+      ],
+      [
+        1152,
+        768
+      ],
+      [
+        1152,
+        1152
+      ],
+      [
+        1152,
+        1536
+      ],
+      [
+        1152,
+        1920
+      ],
+      [
+        1152,
+        2304
+      ],
+      [
+        1536,
+        384
+      ],
+      [
+        1536,
+        768
+      ],
+      [
+        1536,
+        1152
+      ],
+      [
+        1536,
+        1536
+      ],
+      [
+        1536,
+        1920
+      ],
+      [
+        1536,
+        2304
+      ],
+      [
+        1920,
+        384
+      ],
+      [
+        1920,
+        768
+      ],
+      [
+        1920,
+        1152
+      ],
+      [
+        1920,
+        1536
+      ],
+      [
+        1920,
+        1920
+      ],
+      [
+        1920,
+        2304
+      ],
+      [
+        2304,
+        384
+      ],
+      [
+        2304,
+        768
+      ],
+      [
+        2304,
+        1152
+      ],
+      [
+        2304,
+        1536
+      ],
+      [
+        2304,
+        1920
+      ],
+      [
+        2304,
+        2304
+      ]
+    ],
+    "image_mean": [
+      0.48145466,
+      0.4578275,
+      0.40821073
+    ],
+    "image_processor_type": "LlavaOnevisionImageProcessor",
+    "image_std": [
+      0.26862954,
+      0.26130258,
+      0.27577711
+    ],
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "size": {
+      "height": 384,
+      "width": 384
+    }
+  },
+  "image_token": "<image>",
+  "num_image_tokens": 729,
+  "processor_class": "LlavaOnevisionProcessor",
+  "video_processor": {
+    "crop_size": {
+      "height": 384,
+      "width": 384
+    },
+    "do_center_crop": true,
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "do_sample_frames": false,
+    "image_mean": [
+      0.48145466,
+      0.4578275,
+      0.40821073
+    ],
+    "image_std": [
+      0.26862954,
+      0.26130258,
+      0.27577711
+    ],
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "return_metadata": false,
+    "size": {
+      "height": 384,
+      "width": 384
+    },
+    "video_processor_type": "LlavaOnevisionVideoProcessor"
+  },
+  "video_token": "<video>",
+  "vision_aspect_ratio": "anyres_max_9",
+  "vision_feature_select_strategy": "full"
+}

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8f92cf63e0989759370d24108b469c492c12202403f036015307ce49f12cedc
+size 16389

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ed40a0a4e9f365d2c6cc004d97e6705894eba46c8be4c160c1455bc3062dee1
+size 16389

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d688b304d19c260b5cfa471535ed51d7e1d60b3a0d0159dfd1a04b87904a9f42
+size 16389

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9967425ebcaee80d9b518fa0244d52f739b1b983d87cda71d5fede0c073e9d3b
+size 16389

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:469900fd39c667ffbd49c3c407c0ba317a1e9f5f9339a99b5d38423b7d0ce6d4
+size 16389

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:124688471ff2a6e80f2fcefedbf741fb18d08dd539d5bd07a52e81be545142a5
+size 16389

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e69f1ced9f992a72c948698e5eb06088610788988cdb2fdbdd624e064319d60
+size 16389

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a376268a55d6ee10c371c06aa952334c4c6a1af9ea2d71b1951a57367a0c6722
+size 16389

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d6a601258940e988b929d52ce2e72f912eb34df8c86c22458c9543283dfec52c
+size 1593

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9024318c850eaadf26be79389d21b07a7afd8f1af749b89f9109b06c0d12173c
+size 11423018

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/tokenizer_config.json

@@ -0,0 +1,19 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<image>",
+    "<video>"
+  ],
+  "is_local": true,
+  "model_max_length": 1010000,
+  "pad_token": "<|endoftext|>",
+  "processor_class": "LlavaOnevisionProcessor",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9159f6da23bc23ed33f0b207575ec9e1eaf6eadc482c6ea4c4f12d2820fd3112
+size 8977

ckpts/S2/4b/aimv2_large/aimv2_large/v0-20261024-210505/checkpoint-2182/zero_to_fp32.py

@@ -0,0 +1,760 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example:
+#   python zero_to_fp32.py . output_dir/
+#   or
+#   python zero_to_fp32.py . output_dir/ --safe_serialization
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+import gc
+import json
+import numpy as np
+from tqdm import tqdm
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device, weights_only=False)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+    total_files = len(files)
+    state_dicts = []
+    for f in tqdm(files, desc='Loading checkpoint shards'):
+        state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+class GatheredTensor:
+    """
+    A pseudo tensor that collects partitioned weights.
+    It is more memory efficient when there are multiple groups.
+    """
+
+    def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape):
+        self.flat_groups = flat_groups
+        self.flat_groups_offset = flat_groups_offset
+        self.offset = offset
+        self.partitioned_numel = partitioned_numel
+        self.shape = shape
+        self.dtype = self.flat_groups[0][0].dtype
+
+    def contiguous(self):
+        """
+        Merge partitioned weights from flat_groups into a single tensor.
+        """
+        end_idx = self.offset + self.partitioned_numel
+        world_size = len(self.flat_groups)
+        pad_flat_param_chunks = []
+
+        for rank_i in range(world_size):
+            # for each rank, we need to collect weights from related group/groups
+            flat_groups_at_rank_i = self.flat_groups[rank_i]
+            start_group_id = None
+            end_group_id = None
+            for group_id in range(len(self.flat_groups_offset)):
+                if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]:
+                    start_group_id = group_id
+                if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]:
+                    end_group_id = group_id
+                    break
+            # collect weights from related group/groups
+            for group_id in range(start_group_id, end_group_id + 1):
+                flat_tensor = flat_groups_at_rank_i[group_id]
+                start_offset = self.offset - self.flat_groups_offset[group_id]
+                end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id]
+                pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset])
+
+        # collect weights from all ranks
+        pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0)
+        param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous()
+        return param
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size
+
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]]))
+    for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'):
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # memory efficient tensor
+        tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape)
+        state_dict[name] = tensor
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def to_torch_tensor(state_dict, return_empty_tensor=False):
+    """
+    Convert state_dict of GatheredTensor to torch tensor
+    """
+    torch_state_dict = {}
+    converted_tensors = {}
+    for name, tensor in state_dict.items():
+        tensor_id = id(tensor)
+        if tensor_id in converted_tensors:  # shared tensors
+            shared_tensor = torch_state_dict[converted_tensors[tensor_id]]
+            torch_state_dict[name] = shared_tensor
+        else:
+            converted_tensors[tensor_id] = name
+            if return_empty_tensor:
+                torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype)
+            else:
+                torch_state_dict[name] = tensor.contiguous()
+    return torch_state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                             tag=None,
+                                             exclude_frozen_parameters=False,
+                                             lazy_mode=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+        - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient.
+          Convert the pesduo tensor to torch tensor by ``.contiguous()``
+
+    Returns:
+        - pytorch ``state_dict``
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    Note: the above usage may not work if your application doesn't have sufficient free CPU memory.
+    You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint. Or you can load state_dict in lazy mode ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu
+        for name, lazy_tensor in state_dict.item():
+            tensor = lazy_tensor.contiguous()  # to cpu
+            print(name, tensor)
+            # del tensor to release memory if it no longer in use
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+    if lazy_mode:
+        return state_dict
+    else:
+        return to_torch_tensor(state_dict)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
+                                               output_dir,
+                                               max_shard_size="5GB",
+                                               safe_serialization=False,
+                                               tag=None,
+                                               exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_dir``: directory to the pytorch fp32 state_dict output files
+        - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
+        - ``safe_serialization``:  whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+    """
+
+    # Dependency pre-check
+    if safe_serialization:
+        try:
+            from safetensors.torch import save_file
+        except ImportError:
+            print('If you want to use `safe_serialization`, please `pip install safetensors`')
+            raise
+    if max_shard_size is not None:
+        try:
+            from huggingface_hub import split_torch_state_dict_into_shards
+        except ImportError:
+            print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
+            raise
+
+    # Convert zero checkpoint to state_dict
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                                          tag,
+                                                          exclude_frozen_parameters,
+                                                          lazy_mode=True)
+
+    # Shard the model if it is too big.
+    weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
+    if max_shard_size is not None:
+        filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
+        # an memory-efficient approach for sharding
+        empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True)
+        state_dict_split = split_torch_state_dict_into_shards(empty_state_dict,
+                                                              filename_pattern=filename_pattern,
+                                                              max_shard_size=max_shard_size)
+    else:
+        from collections import namedtuple
+        StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
+        state_dict_split = StateDictSplit(is_sharded=False,
+                                          filename_to_tensors={weights_name: list(state_dict.keys())})
+
+    # Save the model by shard
+    os.makedirs(output_dir, exist_ok=True)
+    filename_to_tensors = state_dict_split.filename_to_tensors.items()
+    for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
+        shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors}
+        shard_state_dict = to_torch_tensor(shard_state_dict)
+        output_path = os.path.join(output_dir, shard_file)
+        if safe_serialization:
+            save_file(shard_state_dict, output_path, metadata={"format": "pt"})
+        else:
+            torch.save(shard_state_dict, output_path)
+        # release the memory of current shard
+        for tensor_name in list(shard_state_dict.keys()):
+            del state_dict[tensor_name]
+            del shard_state_dict[tensor_name]
+        del shard_state_dict
+        gc.collect()
+
+    # Save index if sharded
+    if state_dict_split.is_sharded:
+        index = {
+            "metadata": state_dict_split.metadata,
+            "weight_map": state_dict_split.tensor_to_filename,
+        }
+        save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
+        save_index_file = os.path.join(output_dir, save_index_file)
+        with open(save_index_file, "w", encoding="utf-8") as f:
+            content = json.dumps(index, indent=2, sort_keys=True) + "\n"
+            f.write(content)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info("Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info("Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument("output_dir",
+                        type=str,
+                        help="directory to the pytorch fp32 state_dict output files"
+                        "(e.g. path/checkpoint-12-output/)")
+    parser.add_argument(
+        "--max_shard_size",
+        type=str,
+        default="5GB",
+        help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
+        "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
+        "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
+        "without CPU OOM issues.")
+    parser.add_argument(
+        "--safe_serialization",
+        default=False,
+        action='store_true',
+        help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
+                                               args.output_dir,
+                                               max_shard_size=args.max_shard_size,
+                                               safe_serialization=args.safe_serialization,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2023 OpenGVLab
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/args.json

@@ -0,0 +1,394 @@
+{
+  "output_dir": "/home/wangjunjie09/exps/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032",
+  "per_device_train_batch_size": 1,
+  "num_train_epochs": 1.0,
+  "max_steps": -1,
+  "learning_rate": 1e-05,
+  "lr_scheduler_type": "cosine",
+  "lr_scheduler_kwargs": null,
+  "warmup_steps": 0,
+  "optim": "adamw_torch_fused",
+  "optim_args": null,
+  "weight_decay": 0.1,
+  "adam_beta1": 0.9,
+  "adam_beta2": 0.95,
+  "adam_epsilon": 1e-08,
+  "optim_target_modules": null,
+  "gradient_accumulation_steps": 8,
+  "average_tokens_across_devices": true,
+  "max_grad_norm": 1.0,
+  "label_smoothing_factor": 0.0,
+  "bf16": true,
+  "fp16": false,
+  "bf16_full_eval": false,
+  "fp16_full_eval": false,
+  "tf32": null,
+  "gradient_checkpointing": true,
+  "gradient_checkpointing_kwargs": "{\"use_reentrant\": false}",
+  "torch_compile": false,
+  "torch_compile_backend": null,
+  "torch_compile_mode": null,
+  "use_liger_kernel": false,
+  "liger_kernel_config": null,
+  "use_cache": false,
+  "neftune_noise_alpha": null,
+  "torch_empty_cache_steps": null,
+  "auto_find_batch_size": false,
+  "logging_strategy": "steps",
+  "logging_steps": 5,
+  "logging_first_step": true,
+  "log_on_each_node": true,
+  "logging_nan_inf_filter": true,
+  "include_num_input_tokens_seen": false,
+  "log_level": "passive",
+  "log_level_replica": "warning",
+  "disable_tqdm": null,
+  "report_to": [
+    "none"
+  ],
+  "run_name": "/home/wangjunjie09/exps/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032",
+  "project": "huggingface",
+  "trackio_space_id": "trackio",
+  "eval_strategy": "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": null,
+  "include_for_metrics": [],
+  "batch_eval_metrics": false,
+  "save_only_model": false,
+  "save_strategy": "steps",
+  "save_steps": 100.0,
+  "save_on_each_node": false,
+  "save_total_limit": 1,
+  "enable_jit_checkpoint": false,
+  "push_to_hub": false,
+  "hub_token": null,
+  "hub_private_repo": null,
+  "hub_model_id": null,
+  "hub_strategy": "every_save",
+  "hub_always_push": false,
+  "hub_revision": null,
+  "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": "{\"dispatch_batches\": false}",
+  "parallelism_config": null,
+  "dataloader_drop_last": false,
+  "dataloader_num_workers": 8,
+  "dataloader_pin_memory": true,
+  "dataloader_persistent_workers": true,
+  "dataloader_prefetch_factor": 4,
+  "remove_unused_columns": true,
+  "label_names": null,
+  "train_sampling_strategy": "random",
+  "length_column_name": "length",
+  "ddp_find_unused_parameters": null,
+  "ddp_bucket_cap_mb": null,
+  "ddp_broadcast_buffers": null,
+  "ddp_backend": null,
+  "ddp_timeout": 7200,
+  "fsdp": [],
+  "fsdp_config": null,
+  "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": null,
+  "skip_memory_metrics": true,
+  "do_train": false,
+  "do_eval": false,
+  "do_predict": false,
+  "resume_from_checkpoint": null,
+  "warmup_ratio": 0.05,
+  "logging_dir": "/home/wangjunjie09/exps/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/runs",
+  "local_rank": 0,
+  "sortish_sampler": false,
+  "predict_with_generate": false,
+  "generation_max_length": null,
+  "generation_num_beams": null,
+  "generation_config": null,
+  "tuner_backend": "peft",
+  "llm_checkpointing": 16,
+  "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": null,
+  "aligner_lr": null,
+  "vit_lr": 1e-06,
+  "use_logits_to_keep": true,
+  "ds3_gather_for_generation": true,
+  "resume_only_model": false,
+  "optimizer": null,
+  "loss_type": null,
+  "eval_metric": null,
+  "callbacks": [],
+  "early_stop_interval": null,
+  "eval_use_evalscope": false,
+  "eval_dataset": [],
+  "eval_dataset_args": null,
+  "eval_limit": null,
+  "eval_generation_config": null,
+  "extra_eval_args": null,
+  "tuner_type": "full",
+  "use_galore": false,
+  "galore_target_modules": null,
+  "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,
+  "use_ray": false,
+  "ray_exp_name": null,
+  "device_groups": null,
+  "model": "/home/wangjunjie09/exps/S1/4b/internvl3_5_2b/internvit_baseline/v0-20261016-074453/checkpoint-2181",
+  "model_type": "llava_internvit_qwen3",
+  "model_revision": null,
+  "task_type": "causal_lm",
+  "torch_dtype": "bfloat16",
+  "attn_impl": "flash_attn",
+  "experts_impl": null,
+  "new_special_tokens": [],
+  "num_labels": null,
+  "problem_type": null,
+  "rope_scaling": null,
+  "device_map": null,
+  "max_memory": {},
+  "max_model_len": null,
+  "local_repo_path": null,
+  "init_strategy": null,
+  "template": "llava_internvit_qwen3",
+  "system": null,
+  "max_length": 16384,
+  "truncation_strategy": "delete",
+  "max_pixels": null,
+  "agent_template": null,
+  "norm_bbox": null,
+  "use_chat_template": true,
+  "padding_side": "right",
+  "padding_free": true,
+  "loss_scale": "default",
+  "sequence_parallel_size": 1,
+  "template_backend": "swift",
+  "response_prefix": null,
+  "enable_thinking": null,
+  "add_non_thinking_prefix": true,
+  "dataset": [],
+  "val_dataset": [],
+  "cached_dataset": [
+    "/share/m2v_intern/wangjunjie09/datasets/prepared/s2/internvit_baseline/image_20pct/anyres/train",
+    "/share/m2v_intern/wangjunjie09/datasets/prepared/s2/internvit_baseline/video_20pct/32f/train"
+  ],
+  "cached_val_dataset": [],
+  "split_dataset_ratio": 0.0,
+  "dataset_num_proc": 16,
+  "load_from_cache_file": false,
+  "dataset_shuffle": true,
+  "val_dataset_shuffle": false,
+  "streaming": false,
+  "interleave_prob": null,
+  "stopping_strategy": "first_exhausted",
+  "shuffle_buffer_size": 1000,
+  "download_mode": "reuse_dataset_if_exists",
+  "columns": {},
+  "strict": false,
+  "model_name": null,
+  "model_author": null,
+  "custom_dataset_info": [],
+  "quant_method": null,
+  "quant_bits": null,
+  "hqq_axis": null,
+  "bnb_4bit_compute_dtype": "bfloat16",
+  "bnb_4bit_quant_type": "nf4",
+  "bnb_4bit_use_double_quant": true,
+  "bnb_4bit_quant_storage": null,
+  "max_new_tokens": 64,
+  "temperature": 0.0,
+  "top_k": null,
+  "top_p": null,
+  "repetition_penalty": null,
+  "num_beams": 1,
+  "stream": false,
+  "stop_words": [],
+  "logprobs": false,
+  "top_logprobs": null,
+  "structured_outputs_regex": null,
+  "train_type": null,
+  "adapters": [],
+  "external_plugins": [
+    "video_mllm/training/model_plugin_internvit.py",
+    "video_mllm/data/dataset_plugin.py",
+    "/home/wangjunjie09/VisionEncoder/video_mllm/training/log_spacing_plugin.py"
+  ],
+  "custom_register_path": [
+    "/home/wangjunjie09/VisionEncoder/video_mllm/training/log_spacing_plugin.py"
+  ],
+  "model_kwargs": {
+    "internvit_vtc": false,
+    "internvit_resolution": 392,
+    "num_frames": 32,
+    "eval_frames": 32,
+    "s2_use_video_frame_cache": true,
+    "s2_video_frame_cache": "/share/m2v_intern/wangjunjie09/datasets/video_cache/s2/frames/jpeg/392/32"
+  },
+  "video_cache": {},
+  "image_teacher_models": {},
+  "load_args": false,
+  "load_data_args": false,
+  "packing": true,
+  "packing_length": 16384,
+  "packing_num_proc": 1,
+  "lazy_tokenize": false,
+  "use_hf": true,
+  "ignore_args_error": false,
+  "use_swift_lora": false,
+  "freeze_parameters": [],
+  "freeze_parameters_regex": null,
+  "freeze_parameters_ratio": 0.0,
+  "trainable_parameters": [
+    "model.multi_modal_projector"
+  ],
+  "trainable_parameters_regex": null,
+  "freeze_llm": false,
+  "freeze_vit": false,
+  "freeze_aligner": false,
+  "target_modules": [
+    "all-linear"
+  ],
+  "target_regex": null,
+  "target_parameters": null,
+  "modules_to_save": [],
+  "lora_rank": 8,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "lora_bias": "none",
+  "lora_dtype": null,
+  "lorap_lr_ratio": null,
+  "use_rslora": false,
+  "use_dora": false,
+  "lora_ga_batch_size": 2,
+  "lora_ga_iters": 2,
+  "lora_ga_max_length": 1024,
+  "lora_ga_direction": "ArB2r",
+  "lora_ga_scale": "stable",
+  "lora_ga_stable_gamma": 16,
+  "init_weights": true,
+  "fourier_n_frequency": 2000,
+  "fourier_scaling": 300.0,
+  "boft_block_size": 4,
+  "boft_block_num": 0,
+  "boft_n_butterfly_factor": 1,
+  "boft_dropout": 0.0,
+  "vera_rank": 256,
+  "vera_projection_prng_key": 0,
+  "vera_dropout": 0.0,
+  "vera_d_initial": 0.1,
+  "adapter_act": "gelu",
+  "adapter_length": 128,
+  "adalora_target_r": 8,
+  "adalora_init_r": 12,
+  "adalora_tinit": 0,
+  "adalora_tfinal": 0,
+  "adalora_deltaT": 1,
+  "adalora_beta1": 0.85,
+  "adalora_beta2": 0.85,
+  "adalora_orth_reg_weight": 0.5,
+  "llamapro_num_new_blocks": 4,
+  "llamapro_num_groups": null,
+  "reft_layer_key": null,
+  "reft_layers": null,
+  "reft_rank": 4,
+  "reft_intervention_type": "LoreftIntervention",
+  "reft_args": null,
+  "swanlab_token": null,
+  "swanlab_project": "ms-swift",
+  "swanlab_workspace": null,
+  "swanlab_exp_name": null,
+  "swanlab_notification_method": null,
+  "swanlab_webhook_url": null,
+  "swanlab_secret": null,
+  "swanlab_sender_email": null,
+  "swanlab_receiver_email": null,
+  "swanlab_smtp_server": null,
+  "swanlab_smtp_port": null,
+  "swanlab_email_language": "zh",
+  "swanlab_mode": "cloud",
+  "add_version": true,
+  "create_checkpoint_symlink": false,
+  "v12_resolution": 384,
+  "zero_hpz_partition_size": null,
+  "deepspeed_autotp_size": null,
+  "swift_version": "4.1.0.dev0",
+  "ckpt_dir": "/home/wangjunjie09/exps/S1/4b/internvl3_5_2b/internvit_baseline/v0-20261016-074453/checkpoint-2181",
+  "rank": 0,
+  "global_world_size": 8,
+  "local_world_size": 8,
+  "model_suffix": "checkpoint-2181",
+  "model_info": "ModelInfo(model_type='llava_internvit_qwen3', model_dir='/home/wangjunjie09/exps/S1/4b/internvl3_5_2b/internvit_baseline/v0-20261016-074453/checkpoint-2181', torch_dtype=torch.bfloat16, max_model_len=262144, 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)",
+  "model_meta": "ModelMeta(model_type='llava_internvit_qwen3', model_groups=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen3-4B-Instruct-2507', hf_model_id='Qwen/Qwen3-4B-Instruct-2507', model_path=None, ms_revision=None, hf_revision=None)], template=None, ignore_patterns=None, requires=None, tags=[])], loader=<class 'model_plugin_internvit.InternViTLoader'>, template='llava_internvit_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=['LlavaInternViTForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=[], tags=['vision', 'video'])",
+  "model_dir": "/home/wangjunjie09/exps/S1/4b/internvl3_5_2b/internvit_baseline/v0-20261016-074453/checkpoint-2181",
+  "template_meta": "QwenTemplateMeta(template_type='llava_internvit_qwen3', 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 'model_plugin_internvit.InternViTTemplate'>, 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='', history_thinking_prefix='')",
+  "_val_dataset_exists": false,
+  "hub": "<class 'swift.hub.hub.HFHub'>",
+  "evaluation_strategy": "steps",
+  "training_args": "Seq2SeqTrainingArguments(output_dir='/home/wangjunjie09/exps/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032', per_device_train_batch_size=1, num_train_epochs=1.0, max_steps=-1, learning_rate=1e-05, 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/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032', 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/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/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=16, 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=None, 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='full', 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)"
+}

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/chat_template.jinja

@@ -0,0 +1,61 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/chat_template.json

@@ -0,0 +1,3 @@
+{
+  "chat_template": "{% for message in messages %}{{'<|im_start|>' + message['role'] + ' '}}{# Render all images first #}{% for content in message['content'] | selectattr('type', 'equalto', 'image') %}{{ '<image>' }}{% endfor %}{# Render all video then #}{% for content in message['content'] | selectattr('type', 'equalto', 'video') %}{{ '<video>' }}{% endfor %}{# Render all text next #}{% if message['role'] != 'assistant' %}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{{ '\n' + content['text'] }}{% endfor %}{% else %}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{% generation %}{{ '\n' + content['text'] }}{% endgeneration %}{% endfor %}{% endif %}{{'<|im_end|>'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}"
+}

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/config.json

@@ -0,0 +1,298 @@
+{
+  "architectures": [
+    "LlavaInternViTForConditionalGeneration"
+  ],
+  "bos_token_id": null,
+  "dtype": "bfloat16",
+  "eos_token_id": 151645,
+  "eval_num_frames": 32,
+  "hidden_size": 2560,
+  "image_grid_pinpoints": [
+    [
+      392,
+      392
+    ],
+    [
+      392,
+      784
+    ],
+    [
+      392,
+      1176
+    ],
+    [
+      392,
+      1568
+    ],
+    [
+      392,
+      1960
+    ],
+    [
+      392,
+      2352
+    ],
+    [
+      784,
+      392
+    ],
+    [
+      784,
+      784
+    ],
+    [
+      784,
+      1176
+    ],
+    [
+      784,
+      1568
+    ],
+    [
+      784,
+      1960
+    ],
+    [
+      784,
+      2352
+    ],
+    [
+      1176,
+      392
+    ],
+    [
+      1176,
+      784
+    ],
+    [
+      1176,
+      1176
+    ],
+    [
+      1176,
+      1568
+    ],
+    [
+      1176,
+      1960
+    ],
+    [
+      1176,
+      2352
+    ],
+    [
+      1568,
+      392
+    ],
+    [
+      1568,
+      784
+    ],
+    [
+      1568,
+      1176
+    ],
+    [
+      1568,
+      1568
+    ],
+    [
+      1568,
+      1960
+    ],
+    [
+      1568,
+      2352
+    ],
+    [
+      1960,
+      392
+    ],
+    [
+      1960,
+      784
+    ],
+    [
+      1960,
+      1176
+    ],
+    [
+      1960,
+      1568
+    ],
+    [
+      1960,
+      1960
+    ],
+    [
+      1960,
+      2352
+    ],
+    [
+      2352,
+      392
+    ],
+    [
+      2352,
+      784
+    ],
+    [
+      2352,
+      1176
+    ],
+    [
+      2352,
+      1568
+    ],
+    [
+      2352,
+      1960
+    ],
+    [
+      2352,
+      2352
+    ]
+  ],
+  "image_token_index": 151669,
+  "input_resolution": 392,
+  "model_type": "llava_internvit_qwen3",
+  "multimodal_projector_bias": true,
+  "pad_token_id": 151643,
+  "projector_hidden_act": "gelu",
+  "text_config": {
+    "_name_or_path": "/share/m2v_intern_v3/wangjunjie09/model_cache/huggingface/Qwen/Qwen3-4B-Instruct-2507",
+    "architectures": [
+      "Qwen3ForCausalLM"
+    ],
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "bos_token_id": 151643,
+    "dtype": "bfloat16",
+    "eos_token_id": 151645,
+    "head_dim": 128,
+    "hidden_act": "silu",
+    "hidden_size": 2560,
+    "initializer_range": 0.02,
+    "intermediate_size": 9728,
+    "layer_types": [
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention"
+    ],
+    "max_position_embeddings": 262144,
+    "max_window_layers": 36,
+    "model_type": "qwen3",
+    "num_attention_heads": 32,
+    "num_hidden_layers": 36,
+    "num_key_value_heads": 8,
+    "pad_token_id": 151643,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+      "rope_theta": 5000000,
+      "rope_type": "default"
+    },
+    "sliding_window": null,
+    "tie_word_embeddings": true,
+    "use_cache": false,
+    "use_sliding_window": false,
+    "vocab_size": 151936
+  },
+  "tie_word_embeddings": true,
+  "train_num_frames": 32,
+  "transformers_version": "5.5.4",
+  "use_cache": false,
+  "video_token_index": 151670,
+  "vision_aspect_ratio": "anyres_max_9",
+  "vision_config": {
+    "_attn_implementation_autoset": true,
+    "add_cross_attention": false,
+    "architectures": [
+      "InternVisionModel"
+    ],
+    "attention_dropout": 0.0,
+    "auto_map": {
+      "AutoConfig": "configuration_intern_vit.InternVisionConfig",
+      "AutoModel": "modeling_intern_vit.InternVisionModel"
+    },
+    "bos_token_id": null,
+    "cross_attention_hidden_size": null,
+    "decoder_start_token_id": null,
+    "drop_path_rate": 0.0,
+    "dropout": 0.0,
+    "dtype": "bfloat16",
+    "eos_token_id": null,
+    "finetuning_task": null,
+    "hidden_act": "gelu",
+    "hidden_size": 1024,
+    "image_size": 448,
+    "initializer_factor": 1.0,
+    "initializer_range": 0.02,
+    "intermediate_size": 4096,
+    "is_decoder": false,
+    "layer_norm_eps": 1e-06,
+    "model_type": "intern_vit_6b",
+    "norm_type": "layer_norm",
+    "num_attention_heads": 16,
+    "num_channels": 3,
+    "num_hidden_layers": 24,
+    "pad_token_id": 151643,
+    "patch_size": 14,
+    "prefix": null,
+    "pruned_heads": {},
+    "qk_normalization": false,
+    "qkv_bias": true,
+    "sep_token_id": null,
+    "task_specific_params": null,
+    "tf_legacy_loss": false,
+    "tie_encoder_decoder": false,
+    "tie_word_embeddings": true,
+    "tokenizer_class": null,
+    "torchscript": false,
+    "use_bfloat16": false,
+    "use_fa3": true,
+    "use_flash_attn": true
+  },
+  "vision_feature_layer": -1,
+  "vision_feature_select_strategy": "full",
+  "vtc_enabled": false,
+  "vtc_image_temporal_position": true,
+  "auto_map": {
+    "AutoConfig": "modeling_internvit_qwen3.LlavaInternViTConfig",
+    "AutoModel": "modeling_internvit_qwen3.LlavaInternViTForConditionalGeneration",
+    "AutoModelForCausalLM": "modeling_internvit_qwen3.LlavaInternViTForConditionalGeneration",
+    "AutoModelForImageTextToText": "modeling_internvit_qwen3.LlavaInternViTForConditionalGeneration"
+  }
+}

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/configuration_intern_vit.py

@@ -0,0 +1,119 @@
+# --------------------------------------------------------
+# InternVL
+# Copyright (c) 2024 OpenGVLab
+# Licensed under The MIT License [see LICENSE for details]
+# --------------------------------------------------------
+import os
+from typing import Union
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+logger = logging.get_logger(__name__)
+
+
+class InternVisionConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`InternVisionModel`]. It is used to
+    instantiate a vision encoder according to the specified arguments, defining the model architecture.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+    Args:
+        num_channels (`int`, *optional*, defaults to 3):
+            Number of color channels in the input images (e.g., 3 for RGB).
+        patch_size (`int`, *optional*, defaults to 14):
+            The size (resolution) of each patch.
+        image_size (`int`, *optional*, defaults to 224):
+            The size (resolution) of each image.
+        qkv_bias (`bool`, *optional*, defaults to `False`):
+            Whether to add a bias to the queries and values in the self-attention layers.
+        hidden_size (`int`, *optional*, defaults to 3200):
+            Dimensionality of the encoder layers and the pooler layer.
+        num_attention_heads (`int`, *optional*, defaults to 25):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        intermediate_size (`int`, *optional*, defaults to 12800):
+            Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
+        qk_normalization (`bool`, *optional*, defaults to `True`):
+            Whether to normalize the queries and keys in the self-attention layers.
+        num_hidden_layers (`int`, *optional*, defaults to 48):
+            Number of hidden layers in the Transformer encoder.
+        use_flash_attn (`bool`, *optional*, defaults to `True`):
+            Whether to use flash attention mechanism.
+        hidden_act (`str` or `function`, *optional*, defaults to `"gelu"`):
+            The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
+            `"relu"`, `"selu"` and `"gelu_new"` ``"gelu"` are supported.
+        layer_norm_eps (`float`, *optional*, defaults to 1e-6):
+            The epsilon used by the layer normalization layers.
+        dropout (`float`, *optional*, defaults to 0.0):
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
+        drop_path_rate (`float`, *optional*, defaults to 0.0):
+            Dropout rate for stochastic depth.
+        attention_dropout (`float`, *optional*, defaults to 0.0):
+            The dropout ratio for the attention probabilities.
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        initializer_factor (`float`, *optional*, defaults to 0.1):
+            A factor for layer scale.
+    """
+
+    model_type = 'intern_vit_6b'
+
+    def __init__(
+            self,
+            num_channels=3,
+            patch_size=14,
+            image_size=224,
+            qkv_bias=False,
+            hidden_size=3200,
+            num_attention_heads=25,
+            intermediate_size=12800,
+            qk_normalization=True,
+            num_hidden_layers=48,
+            use_flash_attn=True,
+            hidden_act='gelu',
+            norm_type='rms_norm',
+            layer_norm_eps=1e-6,
+            dropout=0.0,
+            drop_path_rate=0.0,
+            attention_dropout=0.0,
+            initializer_range=0.02,
+            initializer_factor=0.1,
+            **kwargs,
+    ):
+        super().__init__(**kwargs)
+
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.dropout = dropout
+        self.drop_path_rate = drop_path_rate
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.num_channels = num_channels
+        self.patch_size = patch_size
+        self.image_size = image_size
+        self.initializer_range = initializer_range
+        self.initializer_factor = initializer_factor
+        self.attention_dropout = attention_dropout
+        self.layer_norm_eps = layer_norm_eps
+        self.hidden_act = hidden_act
+        self.norm_type = norm_type
+        self.qkv_bias = qkv_bias
+        self.qk_normalization = qk_normalization
+        self.use_flash_attn = use_flash_attn
+
+    @classmethod
+    def from_pretrained(cls, pretrained_model_name_or_path: Union[str, os.PathLike], **kwargs) -> 'PretrainedConfig':
+        config_dict, kwargs = cls.get_config_dict(pretrained_model_name_or_path, **kwargs)
+
+        if 'vision_config' in config_dict:
+            config_dict = config_dict['vision_config']
+
+        if 'model_type' in config_dict and hasattr(cls, 'model_type') and config_dict['model_type'] != cls.model_type:
+            logger.warning(
+                f"You are using a model of type {config_dict['model_type']} to instantiate a model of type "
+                f'{cls.model_type}. This is not supported for all configurations of models and can yield errors.'
+            )
+
+        return cls.from_dict(config_dict, **kwargs)

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/generation_config.json

@@ -0,0 +1,12 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 151643,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "transformers_version": "5.5.4",
+  "use_cache": true
+}

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/latest

@@ -0,0 +1 @@
+global_step2206

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/model-00001-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2270e44464f0a0bcb71d1267bfa5e72fc51395e1635edcf28345b0e3d3ae1eb9
+size 4987507408

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/model-00002-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0669f2b7d897b14717fa9bfcd0bf460ded3bb69e7b9993b992e0f3f280f7011a
+size 4461827952

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/model.safetensors.index.json

@@ -0,0 +1,752 @@
+{
+  "metadata": {
+    "total_parameters": 4335663104,
+    "total_size": 9449238528
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00001-of-00002.safetensors",
+    "model.image_newline": "model-00001-of-00002.safetensors",
+    "model.language_model.embed_tokens.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.17.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.17.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.18.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.19.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.20.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.28.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.29.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.30.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.30.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.31.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.32.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.33.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.34.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.35.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.language_model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.language_model.norm.weight": "model-00002-of-00002.safetensors",
+    "model.multi_modal_projector.linear_1.bias": "model-00002-of-00002.safetensors",
+    "model.multi_modal_projector.linear_1.weight": "model-00002-of-00002.safetensors",
+    "model.multi_modal_projector.linear_2.bias": "model-00002-of-00002.safetensors",
+    "model.multi_modal_projector.linear_2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.embeddings.class_embedding": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.embeddings.patch_embedding.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.embeddings.patch_embedding.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.embeddings.position_embedding": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.0.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.0.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.0.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.0.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.0.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.0.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.0.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.0.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.0.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.0.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.0.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.0.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.0.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.0.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.1.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.1.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.1.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.1.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.1.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.1.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.1.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.1.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.1.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.1.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.1.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.1.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.1.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.1.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.10.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.10.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.10.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.10.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.10.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.10.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.10.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.10.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.10.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.10.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.10.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.10.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.10.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.10.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.11.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.11.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.11.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.11.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.11.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.11.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.11.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.11.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.11.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.11.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.11.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.11.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.11.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.11.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.12.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.12.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.12.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.12.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.12.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.12.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.12.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.12.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.12.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.12.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.12.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.12.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.12.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.12.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.13.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.13.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.13.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.13.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.13.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.13.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.13.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.13.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.13.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.13.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.13.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.13.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.13.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.13.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.14.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.14.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.14.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.14.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.14.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.14.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.14.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.14.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.14.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.14.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.14.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.14.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.14.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.14.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.15.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.15.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.15.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.15.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.15.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.15.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.15.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.15.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.15.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.15.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.15.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.15.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.15.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.15.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.16.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.16.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.16.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.16.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.16.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.16.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.16.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.16.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.16.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.16.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.16.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.16.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.16.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.16.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.17.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.17.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.17.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.17.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.17.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.17.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.17.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.17.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.17.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.17.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.17.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.17.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.17.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.17.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.18.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.18.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.18.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.18.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.18.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.18.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.18.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.18.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.18.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.18.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.18.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.18.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.18.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.18.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.19.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.19.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.19.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.19.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.19.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.19.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.19.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.19.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.19.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.19.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.19.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.19.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.19.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.19.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.2.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.2.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.2.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.2.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.2.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.2.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.2.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.2.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.2.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.2.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.2.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.2.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.2.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.2.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.20.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.20.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.20.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.20.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.20.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.20.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.20.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.20.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.20.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.20.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.20.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.20.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.20.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.20.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.21.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.21.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.21.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.21.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.21.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.21.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.21.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.21.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.21.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.21.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.21.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.21.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.21.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.21.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.22.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.22.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.22.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.22.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.22.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.22.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.22.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.22.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.22.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.22.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.22.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.22.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.22.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.22.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.23.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.23.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.23.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.23.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.23.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.23.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.23.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.23.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.23.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.23.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.23.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.23.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.23.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.23.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.3.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.3.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.3.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.3.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.3.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.3.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.3.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.3.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.3.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.3.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.3.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.3.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.3.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.3.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.4.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.4.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.4.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.4.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.4.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.4.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.4.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.4.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.4.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.4.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.4.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.4.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.4.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.4.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.5.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.5.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.5.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.5.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.5.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.5.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.5.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.5.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.5.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.5.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.5.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.5.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.5.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.5.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.6.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.6.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.6.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.6.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.6.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.6.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.6.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.6.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.6.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.6.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.6.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.6.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.6.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.6.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.7.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.7.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.7.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.7.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.7.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.7.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.7.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.7.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.7.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.7.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.7.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.7.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.7.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.7.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.8.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.8.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.8.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.8.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.8.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.8.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.8.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.8.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.8.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.8.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.8.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.8.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.8.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.8.norm2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.9.attn.proj.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.9.attn.proj.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.9.attn.qkv.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.9.attn.qkv.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.9.ls1": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.9.ls2": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.9.mlp.fc1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.9.mlp.fc1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.9.mlp.fc2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.9.mlp.fc2.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.9.norm1.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.9.norm1.weight": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.9.norm2.bias": "model-00002-of-00002.safetensors",
+    "model.vision_tower.vision.encoder.layers.9.norm2.weight": "model-00002-of-00002.safetensors"
+  }
+}

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/modeling_intern_vit.py

@@ -0,0 +1,433 @@
+# --------------------------------------------------------
+# InternVL
+# Copyright (c) 2024 OpenGVLab
+# Licensed under The MIT License [see LICENSE for details]
+# --------------------------------------------------------
+
+from typing import Optional, Tuple, Union
+
+import torch
+import torch.nn.functional as F
+import torch.utils.checkpoint
+from einops import rearrange
+from timm.layers import DropPath
+from torch import nn
+from transformers.activations import ACT2FN
+from transformers.modeling_outputs import (BaseModelOutput,
+                                           BaseModelOutputWithPooling)
+from transformers.modeling_utils import PreTrainedModel
+from transformers.utils import logging
+
+from .configuration_intern_vit import InternVisionConfig
+
+try:
+    from flash_attn.bert_padding import pad_input, unpad_input
+    from flash_attn.flash_attn_interface import \
+        flash_attn_varlen_qkvpacked_func
+    has_flash_attn = True
+except:
+    print('FlashAttention2 is not installed.')
+    has_flash_attn = False
+
+logger = logging.get_logger(__name__)
+
+
+class FlashAttention(nn.Module):
+    """Implement the scaled dot product attention with softmax.
+    Arguments
+    ---------
+        softmax_scale: The temperature to use for the softmax attention.
+                      (default: 1/sqrt(d_keys) where d_keys is computed at
+                      runtime)
+        attention_dropout: The dropout rate to apply to the attention
+                           (default: 0.0)
+    """
+
+    def __init__(self, softmax_scale=None, attention_dropout=0.0, device=None, dtype=None):
+        super().__init__()
+        self.softmax_scale = softmax_scale
+        self.dropout_p = attention_dropout
+
+    def forward(self, qkv, key_padding_mask=None, causal=False, cu_seqlens=None,
+                max_s=None, need_weights=False):
+        """Implements the multihead softmax attention.
+        Arguments
+        ---------
+            qkv: The tensor containing the query, key, and value. (B, S, 3, H, D) if key_padding_mask is None
+                if unpadded: (nnz, 3, h, d)
+            key_padding_mask: a bool tensor of shape (B, S)
+        """
+        assert not need_weights
+        assert qkv.dtype in [torch.float16, torch.bfloat16]
+        assert qkv.is_cuda
+
+        if cu_seqlens is None:
+            batch_size = qkv.shape[0]
+            seqlen = qkv.shape[1]
+            if key_padding_mask is None:
+                qkv = rearrange(qkv, 'b s ... -> (b s) ...')
+                max_s = seqlen
+                cu_seqlens = torch.arange(0, (batch_size + 1) * seqlen, step=seqlen, dtype=torch.int32,
+                                          device=qkv.device)
+                output = flash_attn_varlen_qkvpacked_func(
+                    qkv, cu_seqlens, max_s, self.dropout_p if self.training else 0.0,
+                    softmax_scale=self.softmax_scale, causal=causal
+                )
+                output = rearrange(output, '(b s) ... -> b s ...', b=batch_size)
+            else:
+                nheads = qkv.shape[-2]
+                x = rearrange(qkv, 'b s three h d -> b s (three h d)')
+                x_unpad, indices, cu_seqlens, max_s = unpad_input(x, key_padding_mask)
+                x_unpad = rearrange(x_unpad, 'nnz (three h d) -> nnz three h d', three=3, h=nheads)
+                output_unpad = flash_attn_varlen_qkvpacked_func(
+                    x_unpad, cu_seqlens, max_s, self.dropout_p if self.training else 0.0,
+                    softmax_scale=self.softmax_scale, causal=causal
+                )
+                output = rearrange(pad_input(rearrange(output_unpad, 'nnz h d -> nnz (h d)'),
+                                             indices, batch_size, seqlen),
+                                   'b s (h d) -> b s h d', h=nheads)
+        else:
+            assert max_s is not None
+            output = flash_attn_varlen_qkvpacked_func(
+                qkv, cu_seqlens, max_s, self.dropout_p if self.training else 0.0,
+                softmax_scale=self.softmax_scale, causal=causal
+            )
+
+        return output, None
+
+
+class InternRMSNorm(nn.Module):
+    def __init__(self, hidden_size, eps=1e-6):
+        super().__init__()
+        self.weight = nn.Parameter(torch.ones(hidden_size))
+        self.variance_epsilon = eps
+
+    def forward(self, hidden_states):
+        input_dtype = hidden_states.dtype
+        hidden_states = hidden_states.to(torch.float32)
+        variance = hidden_states.pow(2).mean(-1, keepdim=True)
+        hidden_states = hidden_states * torch.rsqrt(variance + self.variance_epsilon)
+        return self.weight * hidden_states.to(input_dtype)
+
+
+try:
+    from apex.normalization import FusedRMSNorm
+
+    InternRMSNorm = FusedRMSNorm  # noqa
+
+    logger.info('Discovered apex.normalization.FusedRMSNorm - will use it instead of InternRMSNorm')
+except ImportError:
+    # using the normal InternRMSNorm
+    pass
+except Exception:
+    logger.warning('discovered apex but it failed to load, falling back to InternRMSNorm')
+    pass
+
+
+NORM2FN = {
+    'rms_norm': InternRMSNorm,
+    'layer_norm': nn.LayerNorm,
+}
+
+
+class InternVisionEmbeddings(nn.Module):
+    def __init__(self, config: InternVisionConfig):
+        super().__init__()
+        self.config = config
+        self.embed_dim = config.hidden_size
+        self.image_size = config.image_size
+        self.patch_size = config.patch_size
+
+        self.class_embedding = nn.Parameter(
+            torch.randn(1, 1, self.embed_dim),
+        )
+
+        self.patch_embedding = nn.Conv2d(
+            in_channels=3, out_channels=self.embed_dim, kernel_size=self.patch_size, stride=self.patch_size
+        )
+
+        self.num_patches = (self.image_size // self.patch_size) ** 2
+        self.num_positions = self.num_patches + 1
+
+        self.position_embedding = nn.Parameter(torch.randn(1, self.num_positions, self.embed_dim))
+
+    def _get_pos_embed(self, pos_embed, H, W):
+        target_dtype = pos_embed.dtype
+        pos_embed = pos_embed.float().reshape(
+            1, self.image_size // self.patch_size, self.image_size // self.patch_size, -1).permute(0, 3, 1, 2)
+        pos_embed = F.interpolate(pos_embed, size=(H, W), mode='bicubic', align_corners=False). \
+            reshape(1, -1, H * W).permute(0, 2, 1).to(target_dtype)
+        return pos_embed
+
+    def forward(self, pixel_values: torch.FloatTensor) -> torch.Tensor:
+        target_dtype = self.patch_embedding.weight.dtype
+        patch_embeds = self.patch_embedding(pixel_values)  # shape = [*, channel, width, height]
+        batch_size, _, height, width = patch_embeds.shape
+        patch_embeds = patch_embeds.flatten(2).transpose(1, 2)
+        class_embeds = self.class_embedding.expand(batch_size, 1, -1).to(target_dtype)
+        embeddings = torch.cat([class_embeds, patch_embeds], dim=1)
+        position_embedding = torch.cat([
+            self.position_embedding[:, :1, :],
+            self._get_pos_embed(self.position_embedding[:, 1:, :], height, width)
+        ], dim=1)
+        embeddings = embeddings + position_embedding.to(target_dtype)
+        return embeddings
+
+
+class InternAttention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(self, config: InternVisionConfig):
+        super().__init__()
+        self.config = config
+        self.embed_dim = config.hidden_size
+        self.num_heads = config.num_attention_heads
+        self.use_flash_attn = config.use_flash_attn and has_flash_attn
+        if config.use_flash_attn and not has_flash_attn:
+            print('Warning: Flash Attention is not available, use_flash_attn is set to False.')
+        self.head_dim = self.embed_dim // self.num_heads
+        if self.head_dim * self.num_heads != self.embed_dim:
+            raise ValueError(
+                f'embed_dim must be divisible by num_heads (got `embed_dim`: {self.embed_dim} and `num_heads`:'
+                f' {self.num_heads}).'
+            )
+
+        self.scale = self.head_dim ** -0.5
+        self.qkv = nn.Linear(self.embed_dim, 3 * self.embed_dim, bias=config.qkv_bias)
+        self.attn_drop = nn.Dropout(config.attention_dropout)
+        self.proj_drop = nn.Dropout(config.dropout)
+
+        self.qk_normalization = config.qk_normalization
+
+        if self.qk_normalization:
+            self.q_norm = InternRMSNorm(self.embed_dim, eps=config.layer_norm_eps)
+            self.k_norm = InternRMSNorm(self.embed_dim, eps=config.layer_norm_eps)
+
+        if self.use_flash_attn:
+            self.inner_attn = FlashAttention(attention_dropout=config.attention_dropout)
+        self.proj = nn.Linear(self.embed_dim, self.embed_dim)
+
+    def _naive_attn(self, x):
+        B, N, C = x.shape
+        qkv = self.qkv(x).reshape(B, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4)
+        q, k, v = qkv.unbind(0)  # make torchscript happy (cannot use tensor as tuple)
+
+        if self.qk_normalization:
+            B_, H_, N_, D_ = q.shape
+            q = self.q_norm(q.transpose(1, 2).flatten(-2, -1)).view(B_, N_, H_, D_).transpose(1, 2)
+            k = self.k_norm(k.transpose(1, 2).flatten(-2, -1)).view(B_, N_, H_, D_).transpose(1, 2)
+
+        attn = ((q * self.scale) @ k.transpose(-2, -1))
+        attn = attn.softmax(dim=-1)
+        attn = self.attn_drop(attn)
+
+        x = (attn @ v).transpose(1, 2).reshape(B, N, C)
+        x = self.proj(x)
+        x = self.proj_drop(x)
+        return x
+
+    def _flash_attn(self, x, key_padding_mask=None, need_weights=False):
+        qkv = self.qkv(x)
+        qkv = rearrange(qkv, 'b s (three h d) -> b s three h d', three=3, h=self.num_heads)
+
+        if self.qk_normalization:
+            q, k, v = qkv.unbind(2)
+            q = self.q_norm(q.flatten(-2, -1)).view(q.shape)
+            k = self.k_norm(k.flatten(-2, -1)).view(k.shape)
+            qkv = torch.stack([q, k, v], dim=2)
+
+        context, _ = self.inner_attn(
+            qkv, key_padding_mask=key_padding_mask, need_weights=need_weights, causal=False
+        )
+        outs = self.proj(rearrange(context, 'b s h d -> b s (h d)'))
+        outs = self.proj_drop(outs)
+        return outs
+
+    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
+        x = self._naive_attn(hidden_states) if not self.use_flash_attn else self._flash_attn(hidden_states)
+        return x
+
+
+class InternMLP(nn.Module):
+    def __init__(self, config: InternVisionConfig):
+        super().__init__()
+        self.config = config
+        self.act = ACT2FN[config.hidden_act]
+        self.fc1 = nn.Linear(config.hidden_size, config.intermediate_size)
+        self.fc2 = nn.Linear(config.intermediate_size, config.hidden_size)
+
+    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
+        hidden_states = self.fc1(hidden_states)
+        hidden_states = self.act(hidden_states)
+        hidden_states = self.fc2(hidden_states)
+        return hidden_states
+
+
+class InternVisionEncoderLayer(nn.Module):
+    def __init__(self, config: InternVisionConfig, drop_path_rate: float):
+        super().__init__()
+        self.embed_dim = config.hidden_size
+        self.intermediate_size = config.intermediate_size
+        self.norm_type = config.norm_type
+
+        self.attn = InternAttention(config)
+        self.mlp = InternMLP(config)
+        self.norm1 = NORM2FN[self.norm_type](self.embed_dim, eps=config.layer_norm_eps)
+        self.norm2 = NORM2FN[self.norm_type](self.embed_dim, eps=config.layer_norm_eps)
+
+        self.ls1 = nn.Parameter(config.initializer_factor * torch.ones(self.embed_dim))
+        self.ls2 = nn.Parameter(config.initializer_factor * torch.ones(self.embed_dim))
+        self.drop_path1 = DropPath(drop_path_rate) if drop_path_rate > 0. else nn.Identity()
+        self.drop_path2 = DropPath(drop_path_rate) if drop_path_rate > 0. else nn.Identity()
+
+    def forward(
+            self,
+            hidden_states: torch.Tensor,
+    ) -> Tuple[torch.FloatTensor, Optional[torch.FloatTensor], Optional[Tuple[torch.FloatTensor]]]:
+        """
+        Args:
+            hidden_states (`Tuple[torch.FloatTensor, Optional[torch.FloatTensor]]`): input to the layer of shape `(batch, seq_len, embed_dim)`
+        """
+        hidden_states = hidden_states + self.drop_path1(self.attn(self.norm1(hidden_states).to(hidden_states.dtype)) * self.ls1)
+
+        hidden_states = hidden_states + self.drop_path2(self.mlp(self.norm2(hidden_states).to(hidden_states.dtype)) * self.ls2)
+
+        return hidden_states
+
+
+class InternVisionEncoder(nn.Module):
+    """
+    Transformer encoder consisting of `config.num_hidden_layers` self attention layers. Each layer is a
+    [`InternEncoderLayer`].
+
+    Args:
+        config (`InternConfig`):
+            The corresponding vision configuration for the `InternEncoder`.
+    """
+
+    def __init__(self, config: InternVisionConfig):
+        super().__init__()
+        self.config = config
+        # stochastic depth decay rule
+        dpr = [x.item() for x in torch.linspace(0, config.drop_path_rate, config.num_hidden_layers)]
+        self.layers = nn.ModuleList([
+            InternVisionEncoderLayer(config, dpr[idx]) for idx in range(config.num_hidden_layers)])
+        self.gradient_checkpointing = True
+
+    def forward(
+            self,
+            inputs_embeds,
+            output_hidden_states: Optional[bool] = None,
+            return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, BaseModelOutput]:
+        r"""
+        Args:
+            inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`):
+                Embedded representation of the inputs. Should be float, not int tokens.
+            output_hidden_states (`bool`, *optional*):
+                Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
+                for more detail.
+            return_dict (`bool`, *optional*):
+                Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
+        """
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        encoder_states = () if output_hidden_states else None
+        hidden_states = inputs_embeds
+
+        for idx, encoder_layer in enumerate(self.layers):
+            if output_hidden_states:
+                encoder_states = encoder_states + (hidden_states,)
+            if self.gradient_checkpointing and self.training:
+                layer_outputs = torch.utils.checkpoint.checkpoint(
+                    encoder_layer,
+                    hidden_states, use_reentrant=False)
+            else:
+                layer_outputs = encoder_layer(
+                    hidden_states,
+                )
+            hidden_states = layer_outputs
+
+        if output_hidden_states:
+            encoder_states = encoder_states + (hidden_states,)
+
+        if not return_dict:
+            return tuple(v for v in [hidden_states, encoder_states] if v is not None)
+        return BaseModelOutput(
+            last_hidden_state=hidden_states, hidden_states=encoder_states
+        )
+
+
+class InternVisionModel(PreTrainedModel):
+    main_input_name = 'pixel_values'
+    _supports_flash_attn_2 = True
+    supports_gradient_checkpointing = True
+    config_class = InternVisionConfig
+    _no_split_modules = ['InternVisionEncoderLayer']
+    # support transformers 4.51.+
+    _tp_plan = ''
+
+    def __init__(self, config: InternVisionConfig):
+        super().__init__(config)
+        self.config = config
+
+        self.embeddings = InternVisionEmbeddings(config)
+        self.encoder = InternVisionEncoder(config)
+
+    def resize_pos_embeddings(self, old_size, new_size, patch_size):
+        pos_emb = self.embeddings.position_embedding
+        _, num_positions, embed_dim = pos_emb.shape
+        cls_emb = pos_emb[:, :1, :]
+        pos_emb = pos_emb[:, 1:, :].reshape(1, old_size // patch_size, old_size // patch_size, -1).permute(0, 3, 1, 2)
+        pos_emb = F.interpolate(pos_emb.float(), size=new_size // patch_size, mode='bicubic', align_corners=False)
+        pos_emb = pos_emb.to(cls_emb.dtype).reshape(1, embed_dim, -1).permute(0, 2, 1)
+        pos_emb = torch.cat([cls_emb, pos_emb], dim=1)
+        self.embeddings.position_embedding = nn.Parameter(pos_emb)
+        self.embeddings.image_size = new_size
+        logger.info('Resized position embeddings from {} to {}'.format(old_size, new_size))
+
+    def get_input_embeddings(self):
+        return self.embeddings
+
+    def forward(
+            self,
+            pixel_values: Optional[torch.FloatTensor] = None,
+            output_hidden_states: Optional[bool] = None,
+            return_dict: Optional[bool] = None,
+            pixel_embeds: Optional[torch.FloatTensor] = None,
+    ) -> Union[Tuple, BaseModelOutputWithPooling]:
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        if pixel_values is None and pixel_embeds is None:
+            raise ValueError('You have to specify pixel_values or pixel_embeds')
+
+        if pixel_embeds is not None:
+            hidden_states = pixel_embeds
+        else:
+            if len(pixel_values.shape) == 4:
+                hidden_states = self.embeddings(pixel_values)
+            else:
+                raise ValueError(f'wrong pixel_values size: {pixel_values.shape}')
+        encoder_outputs = self.encoder(
+            inputs_embeds=hidden_states,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+        last_hidden_state = encoder_outputs.last_hidden_state
+        pooled_output = last_hidden_state[:, 0, :]
+
+        if not return_dict:
+            return (last_hidden_state, pooled_output) + encoder_outputs[1:]
+
+        return BaseModelOutputWithPooling(
+            last_hidden_state=last_hidden_state,
+            pooler_output=pooled_output,
+            hidden_states=encoder_outputs.hidden_states,
+            attentions=encoder_outputs.attentions,
+        )

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/modeling_internvit_qwen3.py

@@ -0,0 +1,130 @@
+"""InternViT with LLaVA-OV image packing and the Qwen3 language model."""
+
+import math
+from types import SimpleNamespace
+
+import torch
+from torch import nn
+from torch.nn import functional as F
+from transformers import AutoConfig, AutoModel
+from transformers.modeling_outputs import BaseModelOutputWithPooling
+from transformers.models.llava_onevision.configuration_llava_onevision import LlavaOnevisionConfig
+from transformers.models.llava_onevision.modeling_llava_onevision import (
+    LlavaOnevisionModel, LlavaOnevisionForConditionalGeneration, LlavaOnevisionPreTrainedModel,
+    LlavaOnevisionMultiModalProjector, image_size_to_num_patches,
+)
+
+from .configuration_intern_vit import InternVisionConfig
+from .modeling_intern_vit import InternVisionModel
+from .vtc_core import VTCCompressor
+
+AutoConfig.register(InternVisionConfig.model_type, InternVisionConfig, exist_ok=True)
+
+
+class LlavaInternViTConfig(LlavaOnevisionConfig):
+    model_type = "llava_internvit_qwen3"
+
+    def __init__(self, vision_config=None, input_resolution=392, vtc_enabled=False,
+                 vtc_image_temporal_position=True, train_num_frames=64, eval_num_frames=64,
+                 vision_aspect_ratio="anyres_max_9", image_grid_pinpoints=None, **kwargs):
+        if vision_config is None:
+            vision_config = InternVisionConfig()
+        if isinstance(vision_config, dict):
+            vision_config = InternVisionConfig(**vision_config)
+        if image_grid_pinpoints is None:
+            image_grid_pinpoints = [[input_resolution * h, input_resolution * w]
+                                    for h in range(1, 7) for w in range(1, 7)]
+        super().__init__(vision_config=vision_config, vision_aspect_ratio=vision_aspect_ratio,
+                         image_grid_pinpoints=image_grid_pinpoints, **kwargs)
+        self.input_resolution = input_resolution
+        self.vtc_enabled = vtc_enabled
+        self.vtc_image_temporal_position = vtc_image_temporal_position
+        self.train_num_frames = train_num_frames
+        self.eval_num_frames = eval_num_frames
+
+
+class InternViTTower(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        # Upstream InternViT dispatches FA2 via use_flash_attn, outside HF's registry.
+        config.vision_config._attn_implementation = "eager"
+        self.vision = InternVisionModel(config.vision_config)
+        self.vtc_compressor = VTCCompressor(
+            config.vision_config.hidden_size, image_temporal_position=config.vtc_image_temporal_position,
+        ) if config.vtc_enabled else None
+
+    def forward(self, pixels, *, video=False):
+        leading = pixels.shape[:-3]
+        features = self.vision(pixels.flatten(0, -4) if video else pixels, return_dict=True).last_hidden_state[:, 1:]
+        features = features.reshape(*leading, *features.shape[-2:])
+        if self.vtc_compressor is not None:
+            features = self.vtc_compressor(features if video else features.unsqueeze(1), image=not video)
+            if not video:
+                features = features.squeeze(1)
+        return features
+
+
+class LlavaInternViTModel(LlavaOnevisionModel):
+    config_class = LlavaInternViTConfig
+
+    def __init__(self, config):
+        LlavaOnevisionPreTrainedModel.__init__(self, config)
+        self.vision_tower = InternViTTower(config)
+        self.multi_modal_projector = LlavaOnevisionMultiModalProjector(config)
+        self.image_newline = nn.Parameter(torch.randn(config.text_config.hidden_size) / math.sqrt(config.text_config.hidden_size))
+        self.vocab_size = config.text_config.vocab_size
+        self.language_model = AutoModel.from_config(config.text_config)
+        self.post_init()
+
+    def get_image_features(self, pixel_values, image_sizes=None, **kwargs):
+        if self.config.vision_aspect_ratio == "square":
+            pixels = pixel_values[:, 0] if pixel_values.ndim == 5 else pixel_values
+            features = self.multi_modal_projector(self.vision_tower(pixels))
+            newline = self.image_newline[None, None].expand(features.shape[0], 1, -1).to(features)
+            features = torch.cat((features, newline), dim=1)
+            return BaseModelOutputWithPooling(pooler_output=list(features.unbind(0)))
+
+        tile_counts = [image_size_to_num_patches(size, self.config.image_grid_pinpoints,
+                                               self.config.input_resolution) for size in image_sizes]
+        if pixel_values.ndim == 5:
+            pixel_values = torch.cat([pixels[:count] for pixels, count in zip(pixel_values, tile_counts)])
+        features = self.multi_modal_projector(self.vision_tower(pixel_values))
+        features, _ = self.pack_image_features(
+            features.split(tile_counts), image_sizes, image_newline=self.image_newline,
+            vision_aspect_ratio=self.config.vision_aspect_ratio,
+        )
+        return BaseModelOutputWithPooling(pooler_output=features)
+
+    def pack_image_features(self, image_features, image_sizes, image_newline=None, vision_aspect_ratio="anyres_max_9"):
+        # Packing uses the input grid, while the vision weights retain their 448-position grid.
+        geometry = SimpleNamespace(config=SimpleNamespace(
+            vision_config=SimpleNamespace(image_size=self.config.input_resolution,
+                                          patch_size=self.config.vision_config.patch_size),
+            image_grid_pinpoints=self.config.image_grid_pinpoints,
+        ))
+        return LlavaOnevisionModel.pack_image_features(
+            geometry, image_features, image_sizes, image_newline=image_newline,
+            vision_aspect_ratio=vision_aspect_ratio,
+        )
+
+    def get_video_features(self, pixel_values, **kwargs):
+        features = self.multi_modal_projector(self.vision_tower(pixel_values, video=True))
+        batch, frames, patches, width = features.shape
+        side = int(patches ** 0.5)
+        features = features.reshape(batch * frames, side, side, width).permute(0, 3, 1, 2)
+        features = F.interpolate(features, size=(math.ceil(side / 2), math.ceil(side / 2)), mode="bilinear")
+        features = features.permute(0, 2, 3, 1).reshape(batch, -1, width)
+        return BaseModelOutputWithPooling(pooler_output=features)
+
+
+class LlavaInternViTForConditionalGeneration(LlavaOnevisionForConditionalGeneration):
+    config_class = LlavaInternViTConfig
+
+    def __init__(self, config):
+        LlavaOnevisionPreTrainedModel.__init__(self, config)
+        self.model = LlavaInternViTModel(config)
+        self.lm_head = nn.Linear(config.text_config.hidden_size, config.text_config.vocab_size, bias=False)
+        self.post_init()
+
+
+AutoConfig.register(LlavaInternViTConfig.model_type, LlavaInternViTConfig, exist_ok=True)

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/processing_internvit_qwen3.py

@@ -0,0 +1,91 @@
+"""LLaVA-OV AnyRes images and fixed-resolution videos with matching token counts."""
+
+import numpy as np
+import torch
+from PIL import Image
+from transformers import SiglipImageProcessor, LlavaOnevisionImageProcessor, LlavaOnevisionVideoProcessor
+from transformers.feature_extraction_utils import BatchFeature
+from transformers.models.llava_onevision.modeling_llava_onevision import image_size_to_num_patches
+from transformers.models.llava_onevision.processing_llava_onevision import LlavaOnevisionProcessor
+
+
+class LlavaInternViTProcessor(LlavaOnevisionProcessor):
+    image_processor_class = "AutoImageProcessor"
+
+    def __init__(self, image_processor=None, tokenizer=None, video_processor=None,
+                 num_image_tokens=784, vision_feature_select_strategy="full", chat_template=None,
+                 image_token="<image>", video_token="<video>", vision_aspect_ratio="anyres_max_9",
+                 input_resolution=392, patch_size=14, vtc_enabled=False, **kwargs):
+        self.input_resolution = input_resolution
+        self.patch_size = patch_size
+        self.vtc_enabled = vtc_enabled
+        super().__init__(image_processor=image_processor, tokenizer=tokenizer, video_processor=video_processor,
+                         num_image_tokens=num_image_tokens, vision_feature_select_strategy=vision_feature_select_strategy,
+                         chat_template=chat_template, image_token=image_token, video_token=video_token,
+                         vision_aspect_ratio=vision_aspect_ratio, **kwargs)
+        self.frame_processor = SiglipImageProcessor(
+            size={"height": input_resolution, "width": input_resolution},
+            image_mean=image_processor.image_mean, image_std=image_processor.image_std,
+            resample=Image.Resampling.BICUBIC,
+        )
+
+    def video_count(self, frames):
+        return frames // (2 if self.vtc_enabled else 1) * ((self.input_resolution // self.patch_size + 1) // 2) ** 2 + 1
+
+    def _get_number_of_features(self, *args):
+        if self.vision_aspect_ratio == "square":
+            return self.num_image_tokens + 1
+        return super()._get_number_of_features(*args)
+
+    def images(self, images):
+        result = self.image_processor(images, return_tensors="pt")
+        if self.vision_aspect_ratio == "square":
+            result["image_sizes"] = torch.tensor([[self.input_resolution, self.input_resolution]] * len(result["pixel_values"]))
+        else:
+            counts = [image_size_to_num_patches(size, self.image_processor.image_grid_pinpoints,
+                                              self.input_resolution) for size in result["image_sizes"]]
+            # Remove padded tiles before Swift concatenates different images across samples.
+            result["pixel_values"] = torch.cat([pixels[:count] for pixels, count in zip(result["pixel_values"], counts)])
+            result.pop("batch_num_images")
+        return result
+
+    def videos(self, videos):
+        # The same image transform stretches every selected frame once.
+        pixels = []
+        for video in videos:
+            frames = [Image.fromarray(np.asarray(frame)) if not isinstance(frame, Image.Image) else frame for frame in video]
+            pixels.append(self.frame_processor(frames, return_tensors="pt")["pixel_values"])
+        return BatchFeature({"pixel_values_videos": torch.stack(pixels)})
+
+    def __call__(self, images=None, text=None, videos=None, return_tensors=None, **kwargs):
+        texts = [text] if isinstance(text, str) else list(text)
+        media = {}
+        if images is not None:
+            media.update(self.images(images))
+            texts, _ = self._expand_image_tokens(
+                texts, iter(media["image_sizes"]), self.input_resolution, self.input_resolution,
+                self.image_token, iter([1] * len(texts)),
+            )
+        if videos is not None:
+            if isinstance(videos, np.ndarray) and videos.ndim == 4:
+                videos = [videos]
+            media.update(self.videos(videos))
+            count = self.video_count(media["pixel_values_videos"].shape[1])
+            texts = [value.replace(self.video_token, self.video_token * count) for value in texts]
+        tokenizer_kwargs = {key: value for key, value in kwargs.items() if key in ("padding", "truncation", "max_length", "add_special_tokens")}
+        result = dict(self.tokenizer(texts, return_tensors=return_tensors, **tokenizer_kwargs))
+        return BatchFeature({**result, **media})
+
+
+def build_processor(tokenizer, config):
+    image_cls = SiglipImageProcessor if config.vision_aspect_ratio == "square" else LlavaOnevisionImageProcessor
+    image_kwargs = {} if config.vision_aspect_ratio == "square" else {"image_grid_pinpoints": config.image_grid_pinpoints}
+    image = image_cls(size={"height": config.input_resolution, "width": config.input_resolution},
+                      image_mean=[0.485, 0.456, 0.406], image_std=[0.229, 0.224, 0.225],
+                      resample=Image.Resampling.BICUBIC, **image_kwargs)
+    video = LlavaOnevisionVideoProcessor(size={"height": config.input_resolution, "width": config.input_resolution},
+                                        image_mean=image.image_mean, image_std=image.image_std)
+    return LlavaInternViTProcessor(image, tokenizer, video, input_resolution=config.input_resolution,
+                                  num_image_tokens=(config.input_resolution // config.vision_config.patch_size) ** 2,
+                                  patch_size=config.vision_config.patch_size, vtc_enabled=config.vtc_enabled,
+                                  vision_aspect_ratio=config.vision_aspect_ratio)

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/processor_config.json

@@ -0,0 +1,209 @@
+{
+  "image_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_pad": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_grid_pinpoints": [
+      [
+        392,
+        392
+      ],
+      [
+        392,
+        784
+      ],
+      [
+        392,
+        1176
+      ],
+      [
+        392,
+        1568
+      ],
+      [
+        392,
+        1960
+      ],
+      [
+        392,
+        2352
+      ],
+      [
+        784,
+        392
+      ],
+      [
+        784,
+        784
+      ],
+      [
+        784,
+        1176
+      ],
+      [
+        784,
+        1568
+      ],
+      [
+        784,
+        1960
+      ],
+      [
+        784,
+        2352
+      ],
+      [
+        1176,
+        392
+      ],
+      [
+        1176,
+        784
+      ],
+      [
+        1176,
+        1176
+      ],
+      [
+        1176,
+        1568
+      ],
+      [
+        1176,
+        1960
+      ],
+      [
+        1176,
+        2352
+      ],
+      [
+        1568,
+        392
+      ],
+      [
+        1568,
+        784
+      ],
+      [
+        1568,
+        1176
+      ],
+      [
+        1568,
+        1568
+      ],
+      [
+        1568,
+        1960
+      ],
+      [
+        1568,
+        2352
+      ],
+      [
+        1960,
+        392
+      ],
+      [
+        1960,
+        784
+      ],
+      [
+        1960,
+        1176
+      ],
+      [
+        1960,
+        1568
+      ],
+      [
+        1960,
+        1960
+      ],
+      [
+        1960,
+        2352
+      ],
+      [
+        2352,
+        392
+      ],
+      [
+        2352,
+        784
+      ],
+      [
+        2352,
+        1176
+      ],
+      [
+        2352,
+        1568
+      ],
+      [
+        2352,
+        1960
+      ],
+      [
+        2352,
+        2352
+      ]
+    ],
+    "image_mean": [
+      0.485,
+      0.456,
+      0.406
+    ],
+    "image_processor_type": "LlavaOnevisionImageProcessor",
+    "image_std": [
+      0.229,
+      0.224,
+      0.225
+    ],
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "size": {
+      "height": 392,
+      "width": 392
+    }
+  },
+  "image_token": "<image>",
+  "input_resolution": 392,
+  "num_image_tokens": 784,
+  "patch_size": 14,
+  "processor_class": "LlavaInternViTProcessor",
+  "video_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "do_sample_frames": false,
+    "image_mean": [
+      0.485,
+      0.456,
+      0.406
+    ],
+    "image_std": [
+      0.229,
+      0.224,
+      0.225
+    ],
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "return_metadata": false,
+    "size": {
+      "height": 392,
+      "width": 392
+    },
+    "video_processor_type": "LlavaOnevisionVideoProcessor"
+  },
+  "video_token": "<video>",
+  "vision_aspect_ratio": "anyres_max_9",
+  "vision_feature_select_strategy": "full",
+  "vtc_enabled": false,
+  "auto_map": {
+    "AutoProcessor": "processing_internvit_qwen3.LlavaInternViTProcessor"
+  }
+}

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8f92cf63e0989759370d24108b469c492c12202403f036015307ce49f12cedc
+size 16389

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ed40a0a4e9f365d2c6cc004d97e6705894eba46c8be4c160c1455bc3062dee1
+size 16389

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d688b304d19c260b5cfa471535ed51d7e1d60b3a0d0159dfd1a04b87904a9f42
+size 16389

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9967425ebcaee80d9b518fa0244d52f739b1b983d87cda71d5fede0c073e9d3b
+size 16389

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:469900fd39c667ffbd49c3c407c0ba317a1e9f5f9339a99b5d38423b7d0ce6d4
+size 16389

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:124688471ff2a6e80f2fcefedbf741fb18d08dd539d5bd07a52e81be545142a5
+size 16389

ckpts/S2/4b/internvl3_5_2b/internvit_baseline/v0-20261017-034032/checkpoint-2206/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e69f1ced9f992a72c948698e5eb06088610788988cdb2fdbdd624e064319d60
+size 16389