[INFO|2025-03-24 19:58:11] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/config.json [INFO|2025-03-24 19:58:11] configuration_utils.py:771 >> Model config Qwen2_5_VLConfig { "_name_or_path": "Qwen/Qwen2.5-VL-7B-Instruct", "architectures": [ "Qwen2_5_VLForConditionalGeneration" ], "attention_dropout": 0.0, "bos_token_id": 151643, "eos_token_id": 151645, "hidden_act": "silu", "hidden_size": 3584, "image_token_id": 151655, "initializer_range": 0.02, "intermediate_size": 18944, "max_position_embeddings": 128000, "max_window_layers": 28, "model_type": "qwen2_5_vl", "num_attention_heads": 28, "num_hidden_layers": 28, "num_key_value_heads": 4, "rms_norm_eps": 1e-06, "rope_scaling": { "mrope_section": [ 16, 24, 24 ], "rope_type": "default", "type": "default" }, "rope_theta": 1000000.0, "sliding_window": 32768, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.49.0", "use_cache": true, "use_sliding_window": false, "video_token_id": 151656, "vision_config": { "hidden_size": 1280, "in_chans": 3, "model_type": "qwen2_5_vl", "spatial_patch_size": 14, "tokens_per_second": 2 }, "vision_end_token_id": 151653, "vision_start_token_id": 151652, "vision_token_id": 151654, "vocab_size": 152064 } [INFO|2025-03-24 19:58:11] tokenization_utils_base.py:2050 >> loading file vocab.json from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/vocab.json [INFO|2025-03-24 19:58:11] tokenization_utils_base.py:2050 >> loading file merges.txt from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/merges.txt [INFO|2025-03-24 19:58:11] tokenization_utils_base.py:2050 >> loading file tokenizer.json from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/tokenizer.json [INFO|2025-03-24 19:58:11] tokenization_utils_base.py:2050 >> loading file added_tokens.json from cache at None [INFO|2025-03-24 19:58:11] tokenization_utils_base.py:2050 >> loading file special_tokens_map.json from cache at None [INFO|2025-03-24 19:58:11] tokenization_utils_base.py:2050 >> loading file tokenizer_config.json from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/tokenizer_config.json [INFO|2025-03-24 19:58:11] tokenization_utils_base.py:2050 >> loading file chat_template.jinja from cache at None [INFO|2025-03-24 19:58:12] tokenization_utils_base.py:2313 >> Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained. [INFO|2025-03-24 19:58:12] image_processing_base.py:381 >> loading configuration file preprocessor_config.json from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/preprocessor_config.json [INFO|2025-03-24 19:58:12] image_processing_base.py:381 >> loading configuration file preprocessor_config.json from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/preprocessor_config.json [WARNING|2025-03-24 19:58:12] logging.py:329 >> Using a slow image processor as `use_fast` is unset and a slow processor was saved with this model. `use_fast=True` will be the default behavior in v4.48, even if the model was saved with a slow processor. This will result in minor differences in outputs. You'll still be able to use a slow processor with `use_fast=False`. [INFO|2025-03-24 19:58:12] image_processing_base.py:434 >> Image processor Qwen2VLImageProcessor { "do_convert_rgb": true, "do_normalize": true, "do_rescale": true, "do_resize": true, "image_mean": [ 0.48145466, 0.4578275, 0.40821073 ], "image_processor_type": "Qwen2VLImageProcessor", "image_std": [ 0.26862954, 0.26130258, 0.27577711 ], "max_pixels": 12845056, "merge_size": 2, "min_pixels": 3136, "patch_size": 14, "processor_class": "Qwen2_5_VLProcessor", "resample": 3, "rescale_factor": 0.00392156862745098, "size": { "longest_edge": 12845056, "shortest_edge": 3136 }, "temporal_patch_size": 2 } [INFO|2025-03-24 19:58:12] tokenization_utils_base.py:2050 >> loading file vocab.json from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/vocab.json [INFO|2025-03-24 19:58:12] tokenization_utils_base.py:2050 >> loading file merges.txt from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/merges.txt [INFO|2025-03-24 19:58:12] tokenization_utils_base.py:2050 >> loading file tokenizer.json from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/tokenizer.json [INFO|2025-03-24 19:58:12] tokenization_utils_base.py:2050 >> loading file added_tokens.json from cache at None [INFO|2025-03-24 19:58:12] tokenization_utils_base.py:2050 >> loading file special_tokens_map.json from cache at None [INFO|2025-03-24 19:58:12] tokenization_utils_base.py:2050 >> loading file tokenizer_config.json from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/tokenizer_config.json [INFO|2025-03-24 19:58:12] tokenization_utils_base.py:2050 >> loading file chat_template.jinja from cache at None [INFO|2025-03-24 19:58:13] tokenization_utils_base.py:2313 >> Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained. [INFO|2025-03-24 19:58:13] processing_utils.py:876 >> Processor Qwen2_5_VLProcessor: - image_processor: Qwen2VLImageProcessor { "do_convert_rgb": true, "do_normalize": true, "do_rescale": true, "do_resize": true, "image_mean": [ 0.48145466, 0.4578275, 0.40821073 ], "image_processor_type": "Qwen2VLImageProcessor", "image_std": [ 0.26862954, 0.26130258, 0.27577711 ], "max_pixels": 12845056, "merge_size": 2, "min_pixels": 3136, "patch_size": 14, "processor_class": "Qwen2_5_VLProcessor", "resample": 3, "rescale_factor": 0.00392156862745098, "size": { "longest_edge": 12845056, "shortest_edge": 3136 }, "temporal_patch_size": 2 } - tokenizer: Qwen2TokenizerFast(name_or_path='Qwen/Qwen2.5-VL-7B-Instruct', vocab_size=151643, model_max_length=131072, is_fast=True, padding_side='right', truncation_side='right', special_tokens={'eos_token': '<|im_end|>', 'pad_token': '<|endoftext|>', 'additional_special_tokens': ['<|im_start|>', '<|im_end|>', '<|object_ref_start|>', '<|object_ref_end|>', '<|box_start|>', '<|box_end|>', '<|quad_start|>', '<|quad_end|>', '<|vision_start|>', '<|vision_end|>', '<|vision_pad|>', '<|image_pad|>', '<|video_pad|>']}, clean_up_tokenization_spaces=False, added_tokens_decoder={ 151643: AddedToken("<|endoftext|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True), 151644: AddedToken("<|im_start|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True), 151645: AddedToken("<|im_end|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True), 151646: AddedToken("<|object_ref_start|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True), 151647: AddedToken("<|object_ref_end|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True), 151648: AddedToken("<|box_start|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True), 151649: AddedToken("<|box_end|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True), 151650: AddedToken("<|quad_start|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True), 151651: AddedToken("<|quad_end|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True), 151652: AddedToken("<|vision_start|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True), 151653: AddedToken("<|vision_end|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True), 151654: AddedToken("<|vision_pad|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True), 151655: AddedToken("<|image_pad|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True), 151656: AddedToken("<|video_pad|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=True), 151657: AddedToken("", rstrip=False, lstrip=False, single_word=False, normalized=False, special=False), 151658: AddedToken("", rstrip=False, lstrip=False, single_word=False, normalized=False, special=False), 151659: AddedToken("<|fim_prefix|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=False), 151660: AddedToken("<|fim_middle|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=False), 151661: AddedToken("<|fim_suffix|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=False), 151662: AddedToken("<|fim_pad|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=False), 151663: AddedToken("<|repo_name|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=False), 151664: AddedToken("<|file_sep|>", rstrip=False, lstrip=False, single_word=False, normalized=False, special=False), } ) { "processor_class": "Qwen2_5_VLProcessor" } [INFO|2025-03-24 19:58:13] logging.py:157 >> Add <|im_end|> to stop words. [INFO|2025-03-24 19:58:13] logging.py:157 >> Loading dataset BsKIE3.json... [INFO|2025-03-24 19:58:26] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/config.json [INFO|2025-03-24 19:58:26] configuration_utils.py:771 >> Model config Qwen2_5_VLConfig { "_name_or_path": "Qwen/Qwen2.5-VL-7B-Instruct", "architectures": [ "Qwen2_5_VLForConditionalGeneration" ], "attention_dropout": 0.0, "bos_token_id": 151643, "eos_token_id": 151645, "hidden_act": "silu", "hidden_size": 3584, "image_token_id": 151655, "initializer_range": 0.02, "intermediate_size": 18944, "max_position_embeddings": 128000, "max_window_layers": 28, "model_type": "qwen2_5_vl", "num_attention_heads": 28, "num_hidden_layers": 28, "num_key_value_heads": 4, "rms_norm_eps": 1e-06, "rope_scaling": { "mrope_section": [ 16, 24, 24 ], "rope_type": "default", "type": "default" }, "rope_theta": 1000000.0, "sliding_window": 32768, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.49.0", "use_cache": true, "use_sliding_window": false, "video_token_id": 151656, "vision_config": { "hidden_size": 1280, "in_chans": 3, "model_type": "qwen2_5_vl", "spatial_patch_size": 14, "tokens_per_second": 2 }, "vision_end_token_id": 151653, "vision_start_token_id": 151652, "vision_token_id": 151654, "vocab_size": 152064 } [INFO|2025-03-24 19:58:26] modeling_utils.py:3982 >> loading weights file model.safetensors from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/model.safetensors.index.json [INFO|2025-03-24 19:58:26] modeling_utils.py:1633 >> Instantiating Qwen2_5_VLForConditionalGeneration model under default dtype torch.bfloat16. [INFO|2025-03-24 19:58:26] configuration_utils.py:1140 >> Generate config GenerationConfig { "bos_token_id": 151643, "eos_token_id": 151645 } [INFO|2025-03-24 19:58:26] modeling_utils.py:1633 >> Instantiating Qwen2_5_VisionTransformerPretrainedModel model under default dtype torch.bfloat16. [INFO|2025-03-24 19:58:29] modeling_utils.py:4970 >> All model checkpoint weights were used when initializing Qwen2_5_VLForConditionalGeneration. [INFO|2025-03-24 19:58:29] modeling_utils.py:4978 >> All the weights of Qwen2_5_VLForConditionalGeneration were initialized from the model checkpoint at Qwen/Qwen2.5-VL-7B-Instruct. If your task is similar to the task the model of the checkpoint was trained on, you can already use Qwen2_5_VLForConditionalGeneration for predictions without further training. [INFO|2025-03-24 19:58:30] configuration_utils.py:1095 >> loading configuration file generation_config.json from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/generation_config.json [INFO|2025-03-24 19:58:30] configuration_utils.py:1140 >> Generate config GenerationConfig { "bos_token_id": 151643, "do_sample": true, "eos_token_id": [ 151645, 151643 ], "pad_token_id": 151643, "repetition_penalty": 1.05, "temperature": 1e-06 } [INFO|2025-03-24 19:58:30] logging.py:157 >> Gradient checkpointing enabled. [INFO|2025-03-24 19:58:30] logging.py:157 >> Using torch SDPA for faster training and inference. [INFO|2025-03-24 19:58:30] logging.py:157 >> Upcasting trainable params to float32. [INFO|2025-03-24 19:58:30] logging.py:157 >> Fine-tuning method: LoRA [INFO|2025-03-24 19:58:30] logging.py:157 >> Found linear modules: o_proj,v_proj,up_proj,k_proj,q_proj,gate_proj,down_proj [INFO|2025-03-24 19:58:30] logging.py:157 >> Set vision model not trainable: ['visual.patch_embed', 'visual.blocks']. [INFO|2025-03-24 19:58:30] logging.py:157 >> Set multi model projector not trainable: visual.merger. [INFO|2025-03-24 19:58:32] logging.py:157 >> trainable params: 20,185,088 || all params: 8,312,351,744 || trainable%: 0.2428 [INFO|2025-03-24 19:58:32] trainer.py:746 >> Using auto half precision backend [WARNING|2025-03-24 19:58:32] trainer.py:781 >> No label_names provided for model class `PeftModelForCausalLM`. Since `PeftModel` hides base models input arguments, if label_names is not given, label_names can't be set automatically within `Trainer`. Note that empty label_names list will be used instead. [INFO|2025-03-24 19:58:32] trainer.py:2405 >> ***** Running training ***** [INFO|2025-03-24 19:58:32] trainer.py:2406 >> Num examples = 1,050 [INFO|2025-03-24 19:58:32] trainer.py:2407 >> Num Epochs = 3 [INFO|2025-03-24 19:58:32] trainer.py:2408 >> Instantaneous batch size per device = 2 [INFO|2025-03-24 19:58:32] trainer.py:2411 >> Total train batch size (w. parallel, distributed & accumulation) = 16 [INFO|2025-03-24 19:58:32] trainer.py:2412 >> Gradient Accumulation steps = 8 [INFO|2025-03-24 19:58:32] trainer.py:2413 >> Total optimization steps = 195 [INFO|2025-03-24 19:58:32] trainer.py:2414 >> Number of trainable parameters = 20,185,088 [INFO|2025-03-24 20:00:35] logging.py:157 >> {'loss': 0.3150, 'learning_rate': 4.9919e-05, 'epoch': 0.08, 'throughput': 1712.76} [INFO|2025-03-24 20:02:34] logging.py:157 >> {'loss': 0.2410, 'learning_rate': 4.9676e-05, 'epoch': 0.15, 'throughput': 1734.84} [INFO|2025-03-24 20:04:37] logging.py:157 >> {'loss': 0.1709, 'learning_rate': 4.9274e-05, 'epoch': 0.23, 'throughput': 1731.41} [INFO|2025-03-24 20:06:35] logging.py:157 >> {'loss': 0.1225, 'learning_rate': 4.8713e-05, 'epoch': 0.30, 'throughput': 1733.71} [INFO|2025-03-24 20:08:33] logging.py:157 >> {'loss': 0.0945, 'learning_rate': 4.7999e-05, 'epoch': 0.38, 'throughput': 1733.82} [INFO|2025-03-24 20:10:30] logging.py:157 >> {'loss': 0.0871, 'learning_rate': 4.7136e-05, 'epoch': 0.46, 'throughput': 1733.31} [INFO|2025-03-24 20:12:24] logging.py:157 >> {'loss': 0.0634, 'learning_rate': 4.6130e-05, 'epoch': 0.53, 'throughput': 1737.18} [INFO|2025-03-24 20:14:22] logging.py:157 >> {'loss': 0.0783, 'learning_rate': 4.4986e-05, 'epoch': 0.61, 'throughput': 1738.01} [INFO|2025-03-24 20:16:14] logging.py:157 >> {'loss': 0.0632, 'learning_rate': 4.3713e-05, 'epoch': 0.69, 'throughput': 1742.29} [INFO|2025-03-24 20:18:13] logging.py:157 >> {'loss': 0.0713, 'learning_rate': 4.2318e-05, 'epoch': 0.76, 'throughput': 1740.93} [INFO|2025-03-24 20:20:12] logging.py:157 >> {'loss': 0.0589, 'learning_rate': 4.0811e-05, 'epoch': 0.84, 'throughput': 1742.25} [INFO|2025-03-24 20:22:10] logging.py:157 >> {'loss': 0.0635, 'learning_rate': 3.9202e-05, 'epoch': 0.91, 'throughput': 1742.54} [INFO|2025-03-24 20:24:07] logging.py:157 >> {'loss': 0.0583, 'learning_rate': 3.7500e-05, 'epoch': 0.99, 'throughput': 1743.97} [INFO|2025-03-24 20:26:14] logging.py:157 >> {'loss': 0.0686, 'learning_rate': 3.5717e-05, 'epoch': 1.08, 'throughput': 1743.81} [INFO|2025-03-24 20:28:00] logging.py:157 >> {'loss': 0.0538, 'learning_rate': 3.3865e-05, 'epoch': 1.15, 'throughput': 1748.09} [INFO|2025-03-24 20:30:00] logging.py:157 >> {'loss': 0.0546, 'learning_rate': 3.1955e-05, 'epoch': 1.23, 'throughput': 1746.00} [INFO|2025-03-24 20:32:02] logging.py:157 >> {'loss': 0.0447, 'learning_rate': 3.0001e-05, 'epoch': 1.30, 'throughput': 1745.75} [INFO|2025-03-24 20:33:54] logging.py:157 >> {'loss': 0.0536, 'learning_rate': 2.8013e-05, 'epoch': 1.38, 'throughput': 1746.53} [INFO|2025-03-24 20:35:54] logging.py:157 >> {'loss': 0.0489, 'learning_rate': 2.6007e-05, 'epoch': 1.46, 'throughput': 1745.50} [INFO|2025-03-24 20:37:54] logging.py:157 >> {'loss': 0.0441, 'learning_rate': 2.3993e-05, 'epoch': 1.53, 'throughput': 1744.85} [INFO|2025-03-24 20:37:54] trainer.py:3942 >> Saving model checkpoint to saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/checkpoint-100 [INFO|2025-03-24 20:37:54] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/config.json [INFO|2025-03-24 20:37:54] configuration_utils.py:771 >> Model config Qwen2_5_VLConfig { "architectures": [ "Qwen2_5_VLForConditionalGeneration" ], "attention_dropout": 0.0, "bos_token_id": 151643, "eos_token_id": 151645, "hidden_act": "silu", "hidden_size": 3584, "image_token_id": 151655, "initializer_range": 0.02, "intermediate_size": 18944, "max_position_embeddings": 128000, "max_window_layers": 28, "model_type": "qwen2_5_vl", "num_attention_heads": 28, "num_hidden_layers": 28, "num_key_value_heads": 4, "rms_norm_eps": 1e-06, "rope_scaling": { "mrope_section": [ 16, 24, 24 ], "rope_type": "default", "type": "default" }, "rope_theta": 1000000.0, "sliding_window": 32768, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.49.0", "use_cache": true, "use_sliding_window": false, "video_token_id": 151656, "vision_config": { "hidden_size": 1280, "in_chans": 3, "model_type": "qwen2_5_vl", "spatial_patch_size": 14, "tokens_per_second": 2 }, "vision_end_token_id": 151653, "vision_start_token_id": 151652, "vision_token_id": 151654, "vocab_size": 152064 } [INFO|2025-03-24 20:37:54] tokenization_utils_base.py:2500 >> tokenizer config file saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/checkpoint-100/tokenizer_config.json [INFO|2025-03-24 20:37:54] tokenization_utils_base.py:2509 >> Special tokens file saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/checkpoint-100/special_tokens_map.json [INFO|2025-03-24 20:37:55] image_processing_base.py:261 >> Image processor saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/checkpoint-100/preprocessor_config.json [INFO|2025-03-24 20:37:55] tokenization_utils_base.py:2500 >> tokenizer config file saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/checkpoint-100/tokenizer_config.json [INFO|2025-03-24 20:37:55] tokenization_utils_base.py:2509 >> Special tokens file saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/checkpoint-100/special_tokens_map.json [INFO|2025-03-24 20:37:55] processing_utils.py:638 >> chat template saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/checkpoint-100/chat_template.json [INFO|2025-03-24 20:39:53] logging.py:157 >> {'loss': 0.0479, 'learning_rate': 2.1987e-05, 'epoch': 1.61, 'throughput': 1744.13} [INFO|2025-03-24 20:41:55] logging.py:157 >> {'loss': 0.0441, 'learning_rate': 1.9999e-05, 'epoch': 1.69, 'throughput': 1743.47} [INFO|2025-03-24 20:43:53] logging.py:157 >> {'loss': 0.0430, 'learning_rate': 1.8045e-05, 'epoch': 1.76, 'throughput': 1743.40} [INFO|2025-03-24 20:45:54] logging.py:157 >> {'loss': 0.0425, 'learning_rate': 1.6135e-05, 'epoch': 1.84, 'throughput': 1743.59} [INFO|2025-03-24 20:47:45] logging.py:157 >> {'loss': 0.0437, 'learning_rate': 1.4283e-05, 'epoch': 1.91, 'throughput': 1745.66} [INFO|2025-03-24 20:49:35] logging.py:157 >> {'loss': 0.0391, 'learning_rate': 1.2500e-05, 'epoch': 1.99, 'throughput': 1749.15} [INFO|2025-03-24 20:51:26] logging.py:157 >> {'loss': 0.0466, 'learning_rate': 1.0798e-05, 'epoch': 2.08, 'throughput': 1757.58} [INFO|2025-03-24 20:53:02] logging.py:157 >> {'loss': 0.0452, 'learning_rate': 9.1889e-06, 'epoch': 2.15, 'throughput': 1766.25} [INFO|2025-03-24 20:54:49] logging.py:157 >> {'loss': 0.0386, 'learning_rate': 7.6819e-06, 'epoch': 2.23, 'throughput': 1772.54} [INFO|2025-03-24 20:56:27] logging.py:157 >> {'loss': 0.0507, 'learning_rate': 6.2872e-06, 'epoch': 2.30, 'throughput': 1780.01} [INFO|2025-03-24 20:58:08] logging.py:157 >> {'loss': 0.0471, 'learning_rate': 5.0139e-06, 'epoch': 2.38, 'throughput': 1786.96} [INFO|2025-03-24 20:59:49] logging.py:157 >> {'loss': 0.0384, 'learning_rate': 3.8702e-06, 'epoch': 2.46, 'throughput': 1793.35} [INFO|2025-03-24 21:01:32] logging.py:157 >> {'loss': 0.0360, 'learning_rate': 2.8636e-06, 'epoch': 2.53, 'throughput': 1799.46} [INFO|2025-03-24 21:03:13] logging.py:157 >> {'loss': 0.0383, 'learning_rate': 2.0005e-06, 'epoch': 2.61, 'throughput': 1804.98} [INFO|2025-03-24 21:04:55] logging.py:157 >> {'loss': 0.0416, 'learning_rate': 1.2866e-06, 'epoch': 2.69, 'throughput': 1810.00} [INFO|2025-03-24 21:06:41] logging.py:157 >> {'loss': 0.0400, 'learning_rate': 7.2645e-07, 'epoch': 2.76, 'throughput': 1814.13} [INFO|2025-03-24 21:08:33] logging.py:157 >> {'loss': 0.0423, 'learning_rate': 3.2374e-07, 'epoch': 2.84, 'throughput': 1816.96} [INFO|2025-03-24 21:10:15] logging.py:157 >> {'loss': 0.0379, 'learning_rate': 8.1067e-08, 'epoch': 2.91, 'throughput': 1821.78} [INFO|2025-03-24 21:11:54] logging.py:157 >> {'loss': 0.0470, 'learning_rate': 0.0000e+00, 'epoch': 2.99, 'throughput': 1826.28} [INFO|2025-03-24 21:11:54] trainer.py:3942 >> Saving model checkpoint to saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/checkpoint-195 [INFO|2025-03-24 21:11:54] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/config.json [INFO|2025-03-24 21:11:54] configuration_utils.py:771 >> Model config Qwen2_5_VLConfig { "architectures": [ "Qwen2_5_VLForConditionalGeneration" ], "attention_dropout": 0.0, "bos_token_id": 151643, "eos_token_id": 151645, "hidden_act": "silu", "hidden_size": 3584, "image_token_id": 151655, "initializer_range": 0.02, "intermediate_size": 18944, "max_position_embeddings": 128000, "max_window_layers": 28, "model_type": "qwen2_5_vl", "num_attention_heads": 28, "num_hidden_layers": 28, "num_key_value_heads": 4, "rms_norm_eps": 1e-06, "rope_scaling": { "mrope_section": [ 16, 24, 24 ], "rope_type": "default", "type": "default" }, "rope_theta": 1000000.0, "sliding_window": 32768, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.49.0", "use_cache": true, "use_sliding_window": false, "video_token_id": 151656, "vision_config": { "hidden_size": 1280, "in_chans": 3, "model_type": "qwen2_5_vl", "spatial_patch_size": 14, "tokens_per_second": 2 }, "vision_end_token_id": 151653, "vision_start_token_id": 151652, "vision_token_id": 151654, "vocab_size": 152064 } [INFO|2025-03-24 21:11:55] tokenization_utils_base.py:2500 >> tokenizer config file saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/checkpoint-195/tokenizer_config.json [INFO|2025-03-24 21:11:55] tokenization_utils_base.py:2509 >> Special tokens file saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/checkpoint-195/special_tokens_map.json [INFO|2025-03-24 21:11:55] image_processing_base.py:261 >> Image processor saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/checkpoint-195/preprocessor_config.json [INFO|2025-03-24 21:11:55] tokenization_utils_base.py:2500 >> tokenizer config file saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/checkpoint-195/tokenizer_config.json [INFO|2025-03-24 21:11:55] tokenization_utils_base.py:2509 >> Special tokens file saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/checkpoint-195/special_tokens_map.json [INFO|2025-03-24 21:11:55] processing_utils.py:638 >> chat template saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/checkpoint-195/chat_template.json [INFO|2025-03-24 21:11:55] trainer.py:2657 >> Training completed. Do not forget to share your model on huggingface.co/models =) [INFO|2025-03-24 21:11:55] image_processing_base.py:261 >> Image processor saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/preprocessor_config.json [INFO|2025-03-24 21:11:55] tokenization_utils_base.py:2500 >> tokenizer config file saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/tokenizer_config.json [INFO|2025-03-24 21:11:55] tokenization_utils_base.py:2509 >> Special tokens file saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/special_tokens_map.json [INFO|2025-03-24 21:11:56] processing_utils.py:638 >> chat template saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/chat_template.json [INFO|2025-03-24 21:11:56] trainer.py:3942 >> Saving model checkpoint to saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora [INFO|2025-03-24 21:11:56] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--Qwen--Qwen2.5-VL-7B-Instruct/snapshots/5b5eecc7efc2c3e86839993f2689bbbdf06bd8d4/config.json [INFO|2025-03-24 21:11:56] configuration_utils.py:771 >> Model config Qwen2_5_VLConfig { "architectures": [ "Qwen2_5_VLForConditionalGeneration" ], "attention_dropout": 0.0, "bos_token_id": 151643, "eos_token_id": 151645, "hidden_act": "silu", "hidden_size": 3584, "image_token_id": 151655, "initializer_range": 0.02, "intermediate_size": 18944, "max_position_embeddings": 128000, "max_window_layers": 28, "model_type": "qwen2_5_vl", "num_attention_heads": 28, "num_hidden_layers": 28, "num_key_value_heads": 4, "rms_norm_eps": 1e-06, "rope_scaling": { "mrope_section": [ 16, 24, 24 ], "rope_type": "default", "type": "default" }, "rope_theta": 1000000.0, "sliding_window": 32768, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.49.0", "use_cache": true, "use_sliding_window": false, "video_token_id": 151656, "vision_config": { "hidden_size": 1280, "in_chans": 3, "model_type": "qwen2_5_vl", "spatial_patch_size": 14, "tokens_per_second": 2 }, "vision_end_token_id": 151653, "vision_start_token_id": 151652, "vision_token_id": 151654, "vocab_size": 152064 } [INFO|2025-03-24 21:11:56] tokenization_utils_base.py:2500 >> tokenizer config file saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/tokenizer_config.json [INFO|2025-03-24 21:11:56] tokenization_utils_base.py:2509 >> Special tokens file saved in saves/Qwen2.5-VL-7B-Instruct/lora/BS-riche-Qwen2.5-lora/special_tokens_map.json [WARNING|2025-03-24 21:11:56] logging.py:162 >> No metric eval_loss to plot. [WARNING|2025-03-24 21:11:56] logging.py:162 >> No metric eval_accuracy to plot. [INFO|2025-03-24 21:11:56] modelcard.py:449 >> Dropping the following result as it does not have all the necessary fields: {'task': {'name': 'Causal Language Modeling', 'type': 'text-generation'}}