[INFO|2025-02-10 13:06:44] configuration_utils.py:696 >> loading configuration file config.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/config.json [INFO|2025-02-10 13:06:44] configuration_utils.py:768 >> Model config Qwen2VLConfig { "_name_or_path": "Qwen/Qwen2-VL-2B-Instruct-GPTQ-Int4", "architectures": [ "Qwen2VLForConditionalGeneration" ], "attention_dropout": 0.0, "bos_token_id": 151643, "eos_token_id": 151645, "hidden_act": "silu", "hidden_size": 1536, "image_token_id": 151655, "initializer_range": 0.02, "intermediate_size": 8960, "max_position_embeddings": 32768, "max_window_layers": 28, "model_type": "qwen2_vl", "num_attention_heads": 12, "num_hidden_layers": 28, "num_key_value_heads": 2, "quantization_config": { "bits": 4, "damp_percent": 0.1, "dataset": null, "desc_act": false, "group_size": 128, "modules_in_block_to_quantize": null, "quant_method": "gptq", "sym": true, "true_sequential": true }, "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": true, "torch_dtype": "float16", "transformers_version": "4.48.2", "use_cache": true, "use_sliding_window": false, "video_token_id": 151656, "vision_config": { "hidden_size": 1536, "in_chans": 3, "model_type": "qwen2_vl", "spatial_patch_size": 14 }, "vision_end_token_id": 151653, "vision_start_token_id": 151652, "vision_token_id": 151654, "vocab_size": 151936 } [INFO|2025-02-10 13:06:44] tokenization_utils_base.py:2034 >> loading file vocab.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/vocab.json [INFO|2025-02-10 13:06:44] tokenization_utils_base.py:2034 >> loading file merges.txt from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/merges.txt [INFO|2025-02-10 13:06:44] tokenization_utils_base.py:2034 >> loading file tokenizer.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/tokenizer.json [INFO|2025-02-10 13:06:44] tokenization_utils_base.py:2034 >> loading file added_tokens.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/added_tokens.json [INFO|2025-02-10 13:06:44] tokenization_utils_base.py:2034 >> loading file special_tokens_map.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/special_tokens_map.json [INFO|2025-02-10 13:06:44] tokenization_utils_base.py:2034 >> loading file tokenizer_config.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/tokenizer_config.json [INFO|2025-02-10 13:06:44] tokenization_utils_base.py:2034 >> loading file chat_template.jinja from cache at None [INFO|2025-02-10 13:06:45] tokenization_utils_base.py:2304 >> Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained. [INFO|2025-02-10 13:06:45] image_processing_base.py:381 >> loading configuration file preprocessor_config.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/preprocessor_config.json [INFO|2025-02-10 13:06:46] image_processing_base.py:381 >> loading configuration file preprocessor_config.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/preprocessor_config.json [INFO|2025-02-10 13:06:46] 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": 1003520, "merge_size": 2, "min_pixels": 3136, "patch_size": 14, "processor_class": "Qwen2VLProcessor", "resample": 3, "rescale_factor": 0.00392156862745098, "size": { "max_pixels": 1003520, "min_pixels": 3136 }, "temporal_patch_size": 2, "vision_token_id": 151654 } [INFO|2025-02-10 13:06:46] tokenization_utils_base.py:2034 >> loading file vocab.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/vocab.json [INFO|2025-02-10 13:06:46] tokenization_utils_base.py:2034 >> loading file merges.txt from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/merges.txt [INFO|2025-02-10 13:06:46] tokenization_utils_base.py:2034 >> loading file tokenizer.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/tokenizer.json [INFO|2025-02-10 13:06:46] tokenization_utils_base.py:2034 >> loading file added_tokens.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/added_tokens.json [INFO|2025-02-10 13:06:46] tokenization_utils_base.py:2034 >> loading file special_tokens_map.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/special_tokens_map.json [INFO|2025-02-10 13:06:46] tokenization_utils_base.py:2034 >> loading file tokenizer_config.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/tokenizer_config.json [INFO|2025-02-10 13:06:46] tokenization_utils_base.py:2034 >> loading file chat_template.jinja from cache at None [INFO|2025-02-10 13:06:46] tokenization_utils_base.py:2304 >> Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained. [INFO|2025-02-10 13:06:47] processing_utils.py:780 >> Processor Qwen2VLProcessor: - 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": 1003520, "merge_size": 2, "min_pixels": 3136, "patch_size": 14, "processor_class": "Qwen2VLProcessor", "resample": 3, "rescale_factor": 0.00392156862745098, "size": { "max_pixels": 1003520, "min_pixels": 3136 }, "temporal_patch_size": 2, "vision_token_id": 151654 } - tokenizer: Qwen2TokenizerFast(name_or_path='Qwen/Qwen2-VL-2B-Instruct-GPTQ-Int4', vocab_size=151643, model_max_length=32768, is_fast=True, padding_side='left', 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), } ) { "processor_class": "Qwen2VLProcessor" } [INFO|2025-02-10 13:06:47] logging.py:157 >> Add <|im_end|> to stop words. [INFO|2025-02-10 13:06:47] logging.py:157 >> Loading dataset qwen2_vl_dora.json... [INFO|2025-02-10 13:06:48] configuration_utils.py:696 >> loading configuration file config.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/config.json [INFO|2025-02-10 13:06:48] configuration_utils.py:768 >> Model config Qwen2VLConfig { "_name_or_path": "Qwen/Qwen2-VL-2B-Instruct-GPTQ-Int4", "architectures": [ "Qwen2VLForConditionalGeneration" ], "attention_dropout": 0.0, "bos_token_id": 151643, "eos_token_id": 151645, "hidden_act": "silu", "hidden_size": 1536, "image_token_id": 151655, "initializer_range": 0.02, "intermediate_size": 8960, "max_position_embeddings": 32768, "max_window_layers": 28, "model_type": "qwen2_vl", "num_attention_heads": 12, "num_hidden_layers": 28, "num_key_value_heads": 2, "quantization_config": { "bits": 4, "damp_percent": 0.1, "dataset": null, "desc_act": false, "group_size": 128, "modules_in_block_to_quantize": null, "quant_method": "gptq", "sym": true, "true_sequential": true }, "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": true, "torch_dtype": "float16", "transformers_version": "4.48.2", "use_cache": true, "use_sliding_window": false, "video_token_id": 151656, "vision_config": { "hidden_size": 1536, "in_chans": 3, "model_type": "qwen2_vl", "spatial_patch_size": 14 }, "vision_end_token_id": 151653, "vision_start_token_id": 151652, "vision_token_id": 151654, "vocab_size": 151936 } [INFO|2025-02-10 13:06:48] logging.py:157 >> Loading 4-bit GPTQ-quantized model. [INFO|2025-02-10 13:06:48] modeling_utils.py:3904 >> loading weights file model.safetensors from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/model.safetensors [INFO|2025-02-10 13:06:48] modeling_utils.py:1582 >> Instantiating Qwen2VLForConditionalGeneration model under default dtype torch.float16. [INFO|2025-02-10 13:06:48] configuration_utils.py:1140 >> Generate config GenerationConfig { "bos_token_id": 151643, "eos_token_id": 151645 } [INFO|2025-02-10 13:06:48] modeling_utils.py:1582 >> Instantiating Qwen2VisionTransformerPretrainedModel model under default dtype torch.float16. [WARNING|2025-02-10 13:06:48] logging.py:328 >> `Qwen2VLRotaryEmbedding` can now be fully parameterized by passing the model config through the `config` argument. All other arguments will be removed in v4.46 [WARNING|2025-02-10 13:06:48] logging.py:328 >> `loss_type=None` was set in the config but it is unrecognised.Using the default loss: `ForCausalLMLoss`. [INFO|2025-02-10 13:06:50] modeling_utils.py:4888 >> All model checkpoint weights were used when initializing Qwen2VLForConditionalGeneration. [INFO|2025-02-10 13:06:50] modeling_utils.py:4896 >> All the weights of Qwen2VLForConditionalGeneration were initialized from the model checkpoint at Qwen/Qwen2-VL-2B-Instruct-GPTQ-Int4. If your task is similar to the task the model of the checkpoint was trained on, you can already use Qwen2VLForConditionalGeneration for predictions without further training. [INFO|2025-02-10 13:06:50] configuration_utils.py:1095 >> loading configuration file generation_config.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/generation_config.json [INFO|2025-02-10 13:06:50] configuration_utils.py:1140 >> Generate config GenerationConfig { "bos_token_id": 151643, "do_sample": true, "eos_token_id": [ 151645, 151643 ], "pad_token_id": 151643, "temperature": 0.01, "top_k": 1, "top_p": 0.001 } [INFO|2025-02-10 13:06:50] logging.py:157 >> Gradient checkpointing enabled. [INFO|2025-02-10 13:06:50] logging.py:157 >> Casting multimodal projector outputs in torch.float16. [INFO|2025-02-10 13:06:50] logging.py:157 >> Using torch SDPA for faster training and inference. [INFO|2025-02-10 13:06:50] logging.py:157 >> Upcasting trainable params to float32. [INFO|2025-02-10 13:06:50] logging.py:157 >> Fine-tuning method: LoRA [INFO|2025-02-10 13:06:50] logging.py:157 >> Found linear modules: o_proj,k_proj,gate_proj,q_proj,v_proj,down_proj,up_proj [INFO|2025-02-10 13:06:50] logging.py:157 >> Set vision model not trainable: ['visual.patch_embed', 'visual.blocks']. [INFO|2025-02-10 13:06:51] logging.py:157 >> trainable params: 9,232,384 || all params: 907,964,928 || trainable%: 1.0168 [INFO|2025-02-10 13:06:51] trainer.py:741 >> Using auto half precision backend [INFO|2025-02-10 13:06:51] trainer.py:2369 >> ***** Running training ***** [INFO|2025-02-10 13:06:51] trainer.py:2370 >> Num examples = 565 [INFO|2025-02-10 13:06:51] trainer.py:2371 >> Num Epochs = 3 [INFO|2025-02-10 13:06:51] trainer.py:2372 >> Instantaneous batch size per device = 2 [INFO|2025-02-10 13:06:51] trainer.py:2375 >> Total train batch size (w. parallel, distributed & accumulation) = 16 [INFO|2025-02-10 13:06:51] trainer.py:2376 >> Gradient Accumulation steps = 8 [INFO|2025-02-10 13:06:51] trainer.py:2377 >> Total optimization steps = 105 [INFO|2025-02-10 13:06:51] trainer.py:2378 >> Number of trainable parameters = 9,232,384 [INFO|2025-02-10 13:09:19] logging.py:157 >> {'loss': 0.5319, 'learning_rate': 4.9721e-05, 'epoch': 0.14, 'throughput': 602.45} [INFO|2025-02-10 13:11:34] logging.py:157 >> {'loss': 0.3460, 'learning_rate': 4.8889e-05, 'epoch': 0.28, 'throughput': 612.79} [INFO|2025-02-10 13:13:45] logging.py:157 >> {'loss': 0.3586, 'learning_rate': 4.7524e-05, 'epoch': 0.42, 'throughput': 618.32} [INFO|2025-02-10 13:16:24] logging.py:157 >> {'loss': 0.3476, 'learning_rate': 4.5656e-05, 'epoch': 0.57, 'throughput': 618.07} [INFO|2025-02-10 13:18:50] logging.py:157 >> {'loss': 0.2828, 'learning_rate': 4.3326e-05, 'epoch': 0.71, 'throughput': 619.26} [INFO|2025-02-10 13:21:16] logging.py:157 >> {'loss': 0.3261, 'learning_rate': 4.0587e-05, 'epoch': 0.85, 'throughput': 619.90} [INFO|2025-02-10 13:23:37] logging.py:157 >> {'loss': 0.1752, 'learning_rate': 3.7500e-05, 'epoch': 0.99, 'throughput': 618.49} [INFO|2025-02-10 13:25:47] logging.py:157 >> {'loss': 0.1521, 'learning_rate': 3.4134e-05, 'epoch': 1.11, 'throughput': 618.88} [INFO|2025-02-10 13:28:16] logging.py:157 >> {'loss': 0.1056, 'learning_rate': 3.0563e-05, 'epoch': 1.25, 'throughput': 619.19} [INFO|2025-02-10 13:30:37] logging.py:157 >> {'loss': 0.2243, 'learning_rate': 2.6868e-05, 'epoch': 1.40, 'throughput': 616.89} [INFO|2025-02-10 13:30:37] trainer.py:3910 >> Saving model checkpoint to saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-50 [INFO|2025-02-10 13:30:38] configuration_utils.py:696 >> loading configuration file config.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/config.json [INFO|2025-02-10 13:30:38] configuration_utils.py:768 >> Model config Qwen2VLConfig { "_name_or_path": "Qwen/Qwen2-VL-2B-Instruct-319", "architectures": [ "Qwen2VLForConditionalGeneration" ], "attention_dropout": 0.0, "bos_token_id": 151643, "eos_token_id": 151645, "hidden_act": "silu", "hidden_size": 1536, "image_token_id": 151655, "initializer_range": 0.02, "intermediate_size": 8960, "max_position_embeddings": 32768, "max_window_layers": 28, "model_type": "qwen2_vl", "num_attention_heads": 12, "num_hidden_layers": 28, "num_key_value_heads": 2, "quantization_config": { "bits": 4, "damp_percent": 0.1, "dataset": null, "desc_act": false, "group_size": 128, "modules_in_block_to_quantize": null, "quant_method": "gptq", "sym": true, "true_sequential": true }, "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": true, "torch_dtype": "float16", "transformers_version": "4.48.2", "use_cache": true, "use_sliding_window": false, "video_token_id": 151656, "vision_config": { "hidden_size": 1536, "in_chans": 3, "model_type": "qwen2_vl", "spatial_patch_size": 14 }, "vision_end_token_id": 151653, "vision_start_token_id": 151652, "vision_token_id": 151654, "vocab_size": 151936 } [INFO|2025-02-10 13:30:38] tokenization_utils_base.py:2491 >> tokenizer config file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-50/tokenizer_config.json [INFO|2025-02-10 13:30:38] tokenization_utils_base.py:2500 >> Special tokens file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-50/special_tokens_map.json [INFO|2025-02-10 13:30:38] image_processing_base.py:261 >> Image processor saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-50/preprocessor_config.json [INFO|2025-02-10 13:30:38] tokenization_utils_base.py:2491 >> tokenizer config file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-50/tokenizer_config.json [INFO|2025-02-10 13:30:38] tokenization_utils_base.py:2500 >> Special tokens file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-50/special_tokens_map.json [INFO|2025-02-10 13:30:39] processing_utils.py:546 >> chat template saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-50/chat_template.json [INFO|2025-02-10 13:33:01] logging.py:157 >> {'loss': 0.1218, 'learning_rate': 2.3132e-05, 'epoch': 1.54, 'throughput': 617.20} [INFO|2025-02-10 13:35:32] logging.py:157 >> {'loss': 0.1474, 'learning_rate': 1.9437e-05, 'epoch': 1.68, 'throughput': 615.14} [INFO|2025-02-10 13:37:46] logging.py:157 >> {'loss': 0.0559, 'learning_rate': 1.5866e-05, 'epoch': 1.82, 'throughput': 616.26} [INFO|2025-02-10 13:40:12] logging.py:157 >> {'loss': 0.0936, 'learning_rate': 1.2500e-05, 'epoch': 1.96, 'throughput': 616.77} [INFO|2025-02-10 13:42:00] logging.py:157 >> {'loss': 0.0648, 'learning_rate': 9.4128e-06, 'epoch': 2.08, 'throughput': 617.75} [INFO|2025-02-10 13:44:11] logging.py:157 >> {'loss': 0.0676, 'learning_rate': 6.6737e-06, 'epoch': 2.23, 'throughput': 618.56} [INFO|2025-02-10 13:46:37] logging.py:157 >> {'loss': 0.0947, 'learning_rate': 4.3440e-06, 'epoch': 2.37, 'throughput': 618.87} [INFO|2025-02-10 13:49:14] logging.py:157 >> {'loss': 0.0447, 'learning_rate': 2.4758e-06, 'epoch': 2.51, 'throughput': 617.15} [INFO|2025-02-10 13:51:34] logging.py:157 >> {'loss': 0.0535, 'learning_rate': 1.1107e-06, 'epoch': 2.65, 'throughput': 617.57} [INFO|2025-02-10 13:54:02] logging.py:157 >> {'loss': 0.1107, 'learning_rate': 2.7923e-07, 'epoch': 2.79, 'throughput': 617.02} [INFO|2025-02-10 13:54:02] trainer.py:3910 >> Saving model checkpoint to saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-100 [INFO|2025-02-10 13:54:02] configuration_utils.py:696 >> loading configuration file config.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/config.json [INFO|2025-02-10 13:54:02] configuration_utils.py:768 >> Model config Qwen2VLConfig { "_name_or_path": "Qwen/Qwen2-VL-2B-Instruct-319", "architectures": [ "Qwen2VLForConditionalGeneration" ], "attention_dropout": 0.0, "bos_token_id": 151643, "eos_token_id": 151645, "hidden_act": "silu", "hidden_size": 1536, "image_token_id": 151655, "initializer_range": 0.02, "intermediate_size": 8960, "max_position_embeddings": 32768, "max_window_layers": 28, "model_type": "qwen2_vl", "num_attention_heads": 12, "num_hidden_layers": 28, "num_key_value_heads": 2, "quantization_config": { "bits": 4, "damp_percent": 0.1, "dataset": null, "desc_act": false, "group_size": 128, "modules_in_block_to_quantize": null, "quant_method": "gptq", "sym": true, "true_sequential": true }, "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": true, "torch_dtype": "float16", "transformers_version": "4.48.2", "use_cache": true, "use_sliding_window": false, "video_token_id": 151656, "vision_config": { "hidden_size": 1536, "in_chans": 3, "model_type": "qwen2_vl", "spatial_patch_size": 14 }, "vision_end_token_id": 151653, "vision_start_token_id": 151652, "vision_token_id": 151654, "vocab_size": 151936 } [INFO|2025-02-10 13:54:02] tokenization_utils_base.py:2491 >> tokenizer config file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-100/tokenizer_config.json [INFO|2025-02-10 13:54:02] tokenization_utils_base.py:2500 >> Special tokens file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-100/special_tokens_map.json [INFO|2025-02-10 13:54:02] image_processing_base.py:261 >> Image processor saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-100/preprocessor_config.json [INFO|2025-02-10 13:54:02] tokenization_utils_base.py:2491 >> tokenizer config file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-100/tokenizer_config.json [INFO|2025-02-10 13:54:02] tokenization_utils_base.py:2500 >> Special tokens file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-100/special_tokens_map.json [INFO|2025-02-10 13:54:03] processing_utils.py:546 >> chat template saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-100/chat_template.json [INFO|2025-02-10 13:56:39] logging.py:157 >> {'loss': 0.0806, 'learning_rate': 0.0000e+00, 'epoch': 2.93, 'throughput': 616.09} [INFO|2025-02-10 13:56:39] trainer.py:3910 >> Saving model checkpoint to saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-105 [INFO|2025-02-10 13:56:39] configuration_utils.py:696 >> loading configuration file config.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/config.json [INFO|2025-02-10 13:56:39] configuration_utils.py:768 >> Model config Qwen2VLConfig { "_name_or_path": "Qwen/Qwen2-VL-2B-Instruct-319", "architectures": [ "Qwen2VLForConditionalGeneration" ], "attention_dropout": 0.0, "bos_token_id": 151643, "eos_token_id": 151645, "hidden_act": "silu", "hidden_size": 1536, "image_token_id": 151655, "initializer_range": 0.02, "intermediate_size": 8960, "max_position_embeddings": 32768, "max_window_layers": 28, "model_type": "qwen2_vl", "num_attention_heads": 12, "num_hidden_layers": 28, "num_key_value_heads": 2, "quantization_config": { "bits": 4, "damp_percent": 0.1, "dataset": null, "desc_act": false, "group_size": 128, "modules_in_block_to_quantize": null, "quant_method": "gptq", "sym": true, "true_sequential": true }, "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": true, "torch_dtype": "float16", "transformers_version": "4.48.2", "use_cache": true, "use_sliding_window": false, "video_token_id": 151656, "vision_config": { "hidden_size": 1536, "in_chans": 3, "model_type": "qwen2_vl", "spatial_patch_size": 14 }, "vision_end_token_id": 151653, "vision_start_token_id": 151652, "vision_token_id": 151654, "vocab_size": 151936 } [INFO|2025-02-10 13:56:40] tokenization_utils_base.py:2491 >> tokenizer config file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-105/tokenizer_config.json [INFO|2025-02-10 13:56:40] tokenization_utils_base.py:2500 >> Special tokens file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-105/special_tokens_map.json [INFO|2025-02-10 13:56:40] image_processing_base.py:261 >> Image processor saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-105/preprocessor_config.json [INFO|2025-02-10 13:56:40] tokenization_utils_base.py:2491 >> tokenizer config file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-105/tokenizer_config.json [INFO|2025-02-10 13:56:40] tokenization_utils_base.py:2500 >> Special tokens file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-105/special_tokens_map.json [INFO|2025-02-10 13:56:40] processing_utils.py:546 >> chat template saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/checkpoint-105/chat_template.json [INFO|2025-02-10 13:56:40] trainer.py:2643 >> Training completed. Do not forget to share your model on huggingface.co/models =) [INFO|2025-02-10 13:56:40] image_processing_base.py:261 >> Image processor saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/preprocessor_config.json [INFO|2025-02-10 13:56:40] tokenization_utils_base.py:2491 >> tokenizer config file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/tokenizer_config.json [INFO|2025-02-10 13:56:40] tokenization_utils_base.py:2500 >> Special tokens file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/special_tokens_map.json [INFO|2025-02-10 13:56:41] processing_utils.py:546 >> chat template saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/chat_template.json [INFO|2025-02-10 13:56:41] trainer.py:3910 >> Saving model checkpoint to saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4 [INFO|2025-02-10 13:56:41] configuration_utils.py:696 >> loading configuration file config.json from cache at /home/intern01/.cache/huggingface/hub/models--Qwen--Qwen2-VL-2B-Instruct-GPTQ-Int4/snapshots/531c1cb12987e4f4cebdf8df41e08d1c570be0ed/config.json [INFO|2025-02-10 13:56:41] configuration_utils.py:768 >> Model config Qwen2VLConfig { "_name_or_path": "Qwen/Qwen2-VL-2B-Instruct-319", "architectures": [ "Qwen2VLForConditionalGeneration" ], "attention_dropout": 0.0, "bos_token_id": 151643, "eos_token_id": 151645, "hidden_act": "silu", "hidden_size": 1536, "image_token_id": 151655, "initializer_range": 0.02, "intermediate_size": 8960, "max_position_embeddings": 32768, "max_window_layers": 28, "model_type": "qwen2_vl", "num_attention_heads": 12, "num_hidden_layers": 28, "num_key_value_heads": 2, "quantization_config": { "bits": 4, "damp_percent": 0.1, "dataset": null, "desc_act": false, "group_size": 128, "modules_in_block_to_quantize": null, "quant_method": "gptq", "sym": true, "true_sequential": true }, "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": true, "torch_dtype": "float16", "transformers_version": "4.48.2", "use_cache": true, "use_sliding_window": false, "video_token_id": 151656, "vision_config": { "hidden_size": 1536, "in_chans": 3, "model_type": "qwen2_vl", "spatial_patch_size": 14 }, "vision_end_token_id": 151653, "vision_start_token_id": 151652, "vision_token_id": 151654, "vocab_size": 151936 } [INFO|2025-02-10 13:56:42] tokenization_utils_base.py:2491 >> tokenizer config file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/tokenizer_config.json [INFO|2025-02-10 13:56:42] tokenization_utils_base.py:2500 >> Special tokens file saved in saves/Qwen2-VL-2B-Instruct-GPTQ-Int4/lora/250210_Abroad_LoRA_2B_INT4/special_tokens_map.json [WARNING|2025-02-10 13:56:42] logging.py:162 >> No metric eval_loss to plot. [WARNING|2025-02-10 13:56:42] logging.py:162 >> No metric eval_accuracy to plot. [INFO|2025-02-10 13:56:42] modelcard.py:449 >> Dropping the following result as it does not have all the necessary fields: {'task': {'name': 'Causal Language Modeling', 'type': 'text-generation'}}