| [INFO|2025-04-01 09:07:44] tokenization_auto.py:759 >> Could not locate the tokenizer configuration file, will try to use the model config instead. | |
| [INFO|2025-04-01 09:07:44] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 09:07:44] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 09:07:44] tokenization_utils_base.py:2060 >> loading file vocab.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/vocab.json | |
| [INFO|2025-04-01 09:07:44] tokenization_utils_base.py:2060 >> loading file merges.txt from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/merges.txt | |
| [INFO|2025-04-01 09:07:44] tokenization_utils_base.py:2060 >> loading file tokenizer.json from cache at None | |
| [INFO|2025-04-01 09:07:44] tokenization_utils_base.py:2060 >> loading file added_tokens.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/added_tokens.json | |
| [INFO|2025-04-01 09:07:44] tokenization_utils_base.py:2060 >> loading file special_tokens_map.json from cache at None | |
| [INFO|2025-04-01 09:07:44] tokenization_utils_base.py:2060 >> loading file tokenizer_config.json from cache at None | |
| [INFO|2025-04-01 09:07:44] tokenization_utils_base.py:2060 >> loading file chat_template.jinja from cache at None | |
| [INFO|2025-04-01 09:07:44] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 09:07:44] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 09:07:45] tokenization_utils_base.py:2323 >> Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained. | |
| [INFO|2025-04-01 09:07:45] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 09:07:45] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 09:07:45] tokenization_utils_base.py:2323 >> Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained. | |
| [INFO|2025-04-01 09:07:46] image_processing_base.py:381 >> loading configuration file preprocessor_config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/preprocessor_config.json | |
| [INFO|2025-04-01 09:07:46] image_processing_base.py:381 >> loading configuration file preprocessor_config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/preprocessor_config.json | |
| [WARNING|2025-04-01 09:07:46] 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 v4.50, 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|2025-04-01 09:07:46] logging.py:143 >> Loading dataset /content/drive/MyDrive/dataset.jsonl... | |
| [INFO|2025-04-01 09:07:47] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 09:07:47] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 09:07:47] modeling_utils.py:1154 >> loading weights file model.safetensors from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/model.safetensors | |
| [INFO|2025-04-01 09:07:47] modeling_utils.py:2170 >> Instantiating Qwen2VLForConditionalGeneration model under default dtype torch.bfloat16. | |
| [INFO|2025-04-01 09:07:47] configuration_utils.py:1139 >> Generate config GenerationConfig { | |
| "bos_token_id": 151643, | |
| "eos_token_id": 151645, | |
| "pad_token_id": 151654 | |
| } | |
| [INFO|2025-04-01 09:07:47] modeling_utils.py:2170 >> Instantiating Qwen2VisionTransformerPretrainedModel model under default dtype torch.bfloat16. | |
| [INFO|2025-04-01 09:07:50] modeling_utils.py:4987 >> All model checkpoint weights were used when initializing Qwen2VLForConditionalGeneration. | |
| [INFO|2025-04-01 09:07:50] modeling_utils.py:4995 >> All the weights of Qwen2VLForConditionalGeneration were initialized from the model checkpoint at prithivMLmods/Qwen2-VL-OCR-2B-Instruct. | |
| 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|2025-04-01 09:07:50] configuration_utils.py:1094 >> loading configuration file generation_config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/generation_config.json | |
| [INFO|2025-04-01 09:07:50] configuration_utils.py:1139 >> Generate config GenerationConfig { | |
| "bos_token_id": 151643, | |
| "do_sample": true, | |
| "eos_token_id": [ | |
| 151645, | |
| 151643 | |
| ], | |
| "max_length": 32768, | |
| "pad_token_id": 151654, | |
| "temperature": 0.01, | |
| "top_k": 1, | |
| "top_p": 0.001 | |
| } | |
| [INFO|2025-04-01 09:07:50] logging.py:143 >> Gradient checkpointing enabled. | |
| [INFO|2025-04-01 09:07:50] logging.py:143 >> Using torch SDPA for faster training and inference. | |
| [INFO|2025-04-01 09:07:50] logging.py:143 >> Upcasting trainable params to float32. | |
| [INFO|2025-04-01 09:07:50] logging.py:143 >> Fine-tuning method: LoRA | |
| [INFO|2025-04-01 09:07:50] logging.py:143 >> Found linear modules: q_proj,v_proj,gate_proj,down_proj,up_proj,o_proj,k_proj | |
| [INFO|2025-04-01 09:07:50] logging.py:143 >> Set vision model not trainable: ['visual.patch_embed', 'visual.blocks']. | |
| [INFO|2025-04-01 09:07:50] logging.py:143 >> Set multi model projector not trainable: visual.merger. | |
| [INFO|2025-04-01 09:07:51] logging.py:143 >> trainable params: 9,232,384 || all params: 2,218,217,984 || trainable%: 0.4162 | |
| [INFO|2025-04-01 09:07:51] trainer.py:748 >> Using auto half precision backend | |
| [WARNING|2025-04-01 09:07:51] trainer.py:783 >> 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|2025-04-01 09:07:51] trainer.py:2409 >> ***** Running training ***** | |
| [INFO|2025-04-01 09:07:51] trainer.py:2410 >> Num examples = 2,658 | |
| [INFO|2025-04-01 09:07:51] trainer.py:2411 >> Num Epochs = 3 | |
| [INFO|2025-04-01 09:07:51] trainer.py:2412 >> Instantaneous batch size per device = 1 | |
| [INFO|2025-04-01 09:07:51] trainer.py:2415 >> Total train batch size (w. parallel, distributed & accumulation) = 8 | |
| [INFO|2025-04-01 09:07:51] trainer.py:2416 >> Gradient Accumulation steps = 8 | |
| [INFO|2025-04-01 09:07:51] trainer.py:2417 >> Total optimization steps = 996 | |
| [INFO|2025-04-01 09:07:51] trainer.py:2418 >> Number of trainable parameters = 9,232,384 | |
| [INFO|2025-04-01 09:08:45] logging.py:143 >> {'loss': 2.4707, 'learning_rate': 4.9997e-05, 'epoch': 0.02, 'throughput': 108.07} | |
| [INFO|2025-04-01 09:09:39] logging.py:143 >> {'loss': 2.2509, 'learning_rate': 4.9988e-05, 'epoch': 0.03, 'throughput': 106.09} | |
| [INFO|2025-04-01 09:10:31] logging.py:143 >> {'loss': 1.6895, 'learning_rate': 4.9972e-05, 'epoch': 0.05, 'throughput': 106.20} | |
| [INFO|2025-04-01 09:11:24] logging.py:143 >> {'loss': 1.4876, 'learning_rate': 4.9950e-05, 'epoch': 0.06, 'throughput': 107.12} | |
| [INFO|2025-04-01 09:12:19] logging.py:143 >> {'loss': 1.4812, 'learning_rate': 4.9922e-05, 'epoch': 0.08, 'throughput': 108.37} | |
| [INFO|2025-04-01 09:13:11] logging.py:143 >> {'loss': 1.3642, 'learning_rate': 4.9888e-05, 'epoch': 0.09, 'throughput': 108.89} | |
| [INFO|2025-04-01 09:14:06] logging.py:143 >> {'loss': 1.3651, 'learning_rate': 4.9848e-05, 'epoch': 0.11, 'throughput': 109.89} | |
| [INFO|2025-04-01 09:15:00] logging.py:143 >> {'loss': 1.1321, 'learning_rate': 4.9801e-05, 'epoch': 0.12, 'throughput': 110.22} | |
| [INFO|2025-04-01 09:15:53] logging.py:143 >> {'loss': 1.3012, 'learning_rate': 4.9749e-05, 'epoch': 0.14, 'throughput': 110.28} | |
| [INFO|2025-04-01 09:16:46] logging.py:143 >> {'loss': 0.9827, 'learning_rate': 4.9690e-05, 'epoch': 0.15, 'throughput': 110.24} | |
| [INFO|2025-04-01 09:17:40] logging.py:143 >> {'loss': 1.2313, 'learning_rate': 4.9625e-05, 'epoch': 0.17, 'throughput': 110.02} | |
| [INFO|2025-04-01 09:18:35] logging.py:143 >> {'loss': 1.0347, 'learning_rate': 4.9554e-05, 'epoch': 0.18, 'throughput': 110.10} | |
| [INFO|2025-04-01 09:19:28] logging.py:143 >> {'loss': 1.0422, 'learning_rate': 4.9476e-05, 'epoch': 0.20, 'throughput': 110.31} | |
| [INFO|2025-04-01 09:20:22] logging.py:143 >> {'loss': 0.9996, 'learning_rate': 4.9393e-05, 'epoch': 0.21, 'throughput': 110.36} | |
| [INFO|2025-04-01 09:21:16] logging.py:143 >> {'loss': 1.0755, 'learning_rate': 4.9304e-05, 'epoch': 0.23, 'throughput': 110.41} | |
| [INFO|2025-04-01 09:22:07] logging.py:143 >> {'loss': 1.0260, 'learning_rate': 4.9208e-05, 'epoch': 0.24, 'throughput': 110.15} | |
| [INFO|2025-04-01 09:22:59] logging.py:143 >> {'loss': 1.1307, 'learning_rate': 4.9107e-05, 'epoch': 0.26, 'throughput': 110.04} | |
| [INFO|2025-04-01 09:23:52] logging.py:143 >> {'loss': 1.0221, 'learning_rate': 4.8999e-05, 'epoch': 0.27, 'throughput': 109.93} | |
| [INFO|2025-04-01 09:24:43] logging.py:143 >> {'loss': 1.0120, 'learning_rate': 4.8886e-05, 'epoch': 0.29, 'throughput': 109.96} | |
| [INFO|2025-04-01 09:25:34] logging.py:143 >> {'loss': 1.0151, 'learning_rate': 4.8767e-05, 'epoch': 0.30, 'throughput': 109.77} | |
| [INFO|2025-04-01 09:25:34] trainer.py:3966 >> Saving model checkpoint to saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-100 | |
| [INFO|2025-04-01 09:25:34] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 09:25:34] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 09:25:35] tokenization_utils_base.py:2510 >> tokenizer config file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-100/tokenizer_config.json | |
| [INFO|2025-04-01 09:25:35] tokenization_utils_base.py:2519 >> Special tokens file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-100/special_tokens_map.json | |
| [INFO|2025-04-01 09:26:26] logging.py:143 >> {'loss': 1.0028, 'learning_rate': 4.8641e-05, 'epoch': 0.32, 'throughput': 109.30} | |
| [INFO|2025-04-01 09:27:18] logging.py:143 >> {'loss': 1.1430, 'learning_rate': 4.8510e-05, 'epoch': 0.33, 'throughput': 109.20} | |
| [INFO|2025-04-01 09:28:09] logging.py:143 >> {'loss': 0.9695, 'learning_rate': 4.8373e-05, 'epoch': 0.35, 'throughput': 109.18} | |
| [INFO|2025-04-01 09:29:04] logging.py:143 >> {'loss': 0.9017, 'learning_rate': 4.8230e-05, 'epoch': 0.36, 'throughput': 109.34} | |
| [INFO|2025-04-01 09:29:56] logging.py:143 >> {'loss': 1.0350, 'learning_rate': 4.8082e-05, 'epoch': 0.38, 'throughput': 109.29} | |
| [INFO|2025-04-01 09:30:48] logging.py:143 >> {'loss': 1.0128, 'learning_rate': 4.7928e-05, 'epoch': 0.39, 'throughput': 109.25} | |
| [INFO|2025-04-01 09:31:41] logging.py:143 >> {'loss': 0.9432, 'learning_rate': 4.7768e-05, 'epoch': 0.41, 'throughput': 109.32} | |
| [INFO|2025-04-01 09:32:35] logging.py:143 >> {'loss': 1.0344, 'learning_rate': 4.7602e-05, 'epoch': 0.42, 'throughput': 109.50} | |
| [INFO|2025-04-01 09:33:27] logging.py:143 >> {'loss': 0.9452, 'learning_rate': 4.7431e-05, 'epoch': 0.44, 'throughput': 109.55} | |
| [INFO|2025-04-01 09:34:19] logging.py:143 >> {'loss': 0.9559, 'learning_rate': 4.7254e-05, 'epoch': 0.45, 'throughput': 109.53} | |
| [INFO|2025-04-01 09:35:12] logging.py:143 >> {'loss': 0.9726, 'learning_rate': 4.7071e-05, 'epoch': 0.47, 'throughput': 109.60} | |
| [INFO|2025-04-01 09:36:04] logging.py:143 >> {'loss': 0.9344, 'learning_rate': 4.6883e-05, 'epoch': 0.48, 'throughput': 109.48} | |
| [INFO|2025-04-01 09:36:56] logging.py:143 >> {'loss': 0.9497, 'learning_rate': 4.6690e-05, 'epoch': 0.50, 'throughput': 109.40} | |
| [INFO|2025-04-01 09:37:52] logging.py:143 >> {'loss': 1.0570, 'learning_rate': 4.6491e-05, 'epoch': 0.51, 'throughput': 109.59} | |
| [INFO|2025-04-01 09:38:44] logging.py:143 >> {'loss': 0.9847, 'learning_rate': 4.6287e-05, 'epoch': 0.53, 'throughput': 109.65} | |
| [INFO|2025-04-01 09:39:36] logging.py:143 >> {'loss': 1.0010, 'learning_rate': 4.6078e-05, 'epoch': 0.54, 'throughput': 109.52} | |
| [INFO|2025-04-01 09:40:29] logging.py:143 >> {'loss': 0.9384, 'learning_rate': 4.5863e-05, 'epoch': 0.56, 'throughput': 109.56} | |
| [INFO|2025-04-01 09:41:23] logging.py:143 >> {'loss': 1.0312, 'learning_rate': 4.5643e-05, 'epoch': 0.57, 'throughput': 109.68} | |
| [INFO|2025-04-01 09:42:16] logging.py:143 >> {'loss': 0.9112, 'learning_rate': 4.5418e-05, 'epoch': 0.59, 'throughput': 109.70} | |
| [INFO|2025-04-01 09:43:08] logging.py:143 >> {'loss': 0.9967, 'learning_rate': 4.5188e-05, 'epoch': 0.60, 'throughput': 109.66} | |
| [INFO|2025-04-01 09:43:08] trainer.py:3966 >> Saving model checkpoint to saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-200 | |
| [INFO|2025-04-01 09:43:08] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 09:43:08] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 09:43:09] tokenization_utils_base.py:2510 >> tokenizer config file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-200/tokenizer_config.json | |
| [INFO|2025-04-01 09:43:09] tokenization_utils_base.py:2519 >> Special tokens file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-200/special_tokens_map.json | |
| [INFO|2025-04-01 09:44:02] logging.py:143 >> {'loss': 1.0905, 'learning_rate': 4.4953e-05, 'epoch': 0.62, 'throughput': 109.64} | |
| [INFO|2025-04-01 09:44:55] logging.py:143 >> {'loss': 0.9487, 'learning_rate': 4.4713e-05, 'epoch': 0.63, 'throughput': 109.56} | |
| [INFO|2025-04-01 09:45:47] logging.py:143 >> {'loss': 0.8675, 'learning_rate': 4.4468e-05, 'epoch': 0.65, 'throughput': 109.56} | |
| [INFO|2025-04-01 09:46:40] logging.py:143 >> {'loss': 0.8624, 'learning_rate': 4.4219e-05, 'epoch': 0.66, 'throughput': 109.61} | |
| [INFO|2025-04-01 09:47:35] logging.py:143 >> {'loss': 1.0489, 'learning_rate': 4.3964e-05, 'epoch': 0.68, 'throughput': 109.74} | |
| [INFO|2025-04-01 09:48:27] logging.py:143 >> {'loss': 0.9139, 'learning_rate': 4.3705e-05, 'epoch': 0.69, 'throughput': 109.68} | |
| [INFO|2025-04-01 09:49:19] logging.py:143 >> {'loss': 0.9905, 'learning_rate': 4.3441e-05, 'epoch': 0.71, 'throughput': 109.62} | |
| [INFO|2025-04-01 09:50:13] logging.py:143 >> {'loss': 0.8974, 'learning_rate': 4.3172e-05, 'epoch': 0.72, 'throughput': 109.66} | |
| [INFO|2025-04-01 09:51:06] logging.py:143 >> {'loss': 0.9990, 'learning_rate': 4.2899e-05, 'epoch': 0.74, 'throughput': 109.68} | |
| [INFO|2025-04-01 09:51:59] logging.py:143 >> {'loss': 0.9916, 'learning_rate': 4.2622e-05, 'epoch': 0.75, 'throughput': 109.66} | |
| [INFO|2025-04-01 09:52:51] logging.py:143 >> {'loss': 0.9242, 'learning_rate': 4.2340e-05, 'epoch': 0.77, 'throughput': 109.54} | |
| [INFO|2025-04-01 09:53:45] logging.py:143 >> {'loss': 1.0426, 'learning_rate': 4.2054e-05, 'epoch': 0.78, 'throughput': 109.56} | |
| [INFO|2025-04-01 09:54:37] logging.py:143 >> {'loss': 0.8625, 'learning_rate': 4.1763e-05, 'epoch': 0.80, 'throughput': 109.50} | |
| [INFO|2025-04-01 09:55:30] logging.py:143 >> {'loss': 0.9959, 'learning_rate': 4.1469e-05, 'epoch': 0.81, 'throughput': 109.51} | |
| [INFO|2025-04-01 09:56:23] logging.py:143 >> {'loss': 0.9390, 'learning_rate': 4.1170e-05, 'epoch': 0.83, 'throughput': 109.57} | |
| [INFO|2025-04-01 09:57:16] logging.py:143 >> {'loss': 0.9741, 'learning_rate': 4.0867e-05, 'epoch': 0.84, 'throughput': 109.61} | |
| [INFO|2025-04-01 09:58:10] logging.py:143 >> {'loss': 0.9800, 'learning_rate': 4.0561e-05, 'epoch': 0.86, 'throughput': 109.62} | |
| [INFO|2025-04-01 09:59:02] logging.py:143 >> {'loss': 0.8898, 'learning_rate': 4.0250e-05, 'epoch': 0.87, 'throughput': 109.54} | |
| [INFO|2025-04-01 09:59:57] logging.py:143 >> {'loss': 0.9530, 'learning_rate': 3.9936e-05, 'epoch': 0.89, 'throughput': 109.67} | |
| [INFO|2025-04-01 10:00:50] logging.py:143 >> {'loss': 0.9311, 'learning_rate': 3.9618e-05, 'epoch': 0.90, 'throughput': 109.75} | |
| [INFO|2025-04-01 10:00:50] trainer.py:3966 >> Saving model checkpoint to saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-300 | |
| [INFO|2025-04-01 10:00:51] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 10:00:51] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 10:00:51] tokenization_utils_base.py:2510 >> tokenizer config file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-300/tokenizer_config.json | |
| [INFO|2025-04-01 10:00:51] tokenization_utils_base.py:2519 >> Special tokens file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-300/special_tokens_map.json | |
| [INFO|2025-04-01 10:01:43] logging.py:143 >> {'loss': 0.9114, 'learning_rate': 3.9296e-05, 'epoch': 0.92, 'throughput': 109.73} | |
| [INFO|2025-04-01 10:02:37] logging.py:143 >> {'loss': 0.9674, 'learning_rate': 3.8971e-05, 'epoch': 0.93, 'throughput': 109.85} | |
| [INFO|2025-04-01 10:03:30] logging.py:143 >> {'loss': 0.9582, 'learning_rate': 3.8642e-05, 'epoch': 0.95, 'throughput': 109.86} | |
| [INFO|2025-04-01 10:04:22] logging.py:143 >> {'loss': 0.9863, 'learning_rate': 3.8310e-05, 'epoch': 0.96, 'throughput': 109.84} | |
| [INFO|2025-04-01 10:05:16] logging.py:143 >> {'loss': 0.9060, 'learning_rate': 3.7975e-05, 'epoch': 0.98, 'throughput': 109.89} | |
| [INFO|2025-04-01 10:06:09] logging.py:143 >> {'loss': 0.8958, 'learning_rate': 3.7636e-05, 'epoch': 0.99, 'throughput': 109.87} | |
| [INFO|2025-04-01 10:06:54] logging.py:143 >> {'loss': 0.8349, 'learning_rate': 3.7295e-05, 'epoch': 1.01, 'throughput': 109.89} | |
| [INFO|2025-04-01 10:07:46] logging.py:143 >> {'loss': 0.8507, 'learning_rate': 3.6950e-05, 'epoch': 1.02, 'throughput': 109.86} | |
| [INFO|2025-04-01 10:08:39] logging.py:143 >> {'loss': 0.9287, 'learning_rate': 3.6602e-05, 'epoch': 1.04, 'throughput': 109.88} | |
| [INFO|2025-04-01 10:09:32] logging.py:143 >> {'loss': 0.9107, 'learning_rate': 3.6251e-05, 'epoch': 1.05, 'throughput': 109.91} | |
| [INFO|2025-04-01 10:10:25] logging.py:143 >> {'loss': 0.9520, 'learning_rate': 3.5898e-05, 'epoch': 1.07, 'throughput': 109.93} | |
| [INFO|2025-04-01 10:11:18] logging.py:143 >> {'loss': 0.9526, 'learning_rate': 3.5542e-05, 'epoch': 1.08, 'throughput': 109.92} | |
| [INFO|2025-04-01 10:12:11] logging.py:143 >> {'loss': 0.8775, 'learning_rate': 3.5183e-05, 'epoch': 1.10, 'throughput': 109.90} | |
| [INFO|2025-04-01 10:13:05] logging.py:143 >> {'loss': 0.8830, 'learning_rate': 3.4821e-05, 'epoch': 1.11, 'throughput': 109.94} | |
| [INFO|2025-04-01 10:13:58] logging.py:143 >> {'loss': 1.0032, 'learning_rate': 3.4458e-05, 'epoch': 1.13, 'throughput': 109.93} | |
| [INFO|2025-04-01 10:14:51] logging.py:143 >> {'loss': 0.9430, 'learning_rate': 3.4092e-05, 'epoch': 1.14, 'throughput': 109.94} | |
| [INFO|2025-04-01 10:15:43] logging.py:143 >> {'loss': 0.8010, 'learning_rate': 3.3723e-05, 'epoch': 1.16, 'throughput': 109.91} | |
| [INFO|2025-04-01 10:16:35] logging.py:143 >> {'loss': 0.9294, 'learning_rate': 3.3353e-05, 'epoch': 1.17, 'throughput': 109.91} | |
| [INFO|2025-04-01 10:17:28] logging.py:143 >> {'loss': 0.9528, 'learning_rate': 3.2980e-05, 'epoch': 1.19, 'throughput': 109.96} | |
| [INFO|2025-04-01 10:18:20] logging.py:143 >> {'loss': 0.8981, 'learning_rate': 3.2605e-05, 'epoch': 1.20, 'throughput': 109.93} | |
| [INFO|2025-04-01 10:18:21] trainer.py:3966 >> Saving model checkpoint to saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-400 | |
| [INFO|2025-04-01 10:18:21] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 10:18:21] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 10:18:21] tokenization_utils_base.py:2510 >> tokenizer config file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-400/tokenizer_config.json | |
| [INFO|2025-04-01 10:18:21] tokenization_utils_base.py:2519 >> Special tokens file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-400/special_tokens_map.json | |
| [INFO|2025-04-01 10:19:15] logging.py:143 >> {'loss': 0.9823, 'learning_rate': 3.2229e-05, 'epoch': 1.22, 'throughput': 109.94} | |
| [INFO|2025-04-01 10:20:06] logging.py:143 >> {'loss': 0.9047, 'learning_rate': 3.1850e-05, 'epoch': 1.23, 'throughput': 109.86} | |
| [INFO|2025-04-01 10:20:58] logging.py:143 >> {'loss': 0.8582, 'learning_rate': 3.1470e-05, 'epoch': 1.25, 'throughput': 109.84} | |
| [INFO|2025-04-01 10:21:50] logging.py:143 >> {'loss': 0.8787, 'learning_rate': 3.1089e-05, 'epoch': 1.26, 'throughput': 109.85} | |
| [INFO|2025-04-01 10:22:42] logging.py:143 >> {'loss': 0.8729, 'learning_rate': 3.0706e-05, 'epoch': 1.28, 'throughput': 109.85} | |
| [INFO|2025-04-01 10:23:37] logging.py:143 >> {'loss': 0.8772, 'learning_rate': 3.0321e-05, 'epoch': 1.29, 'throughput': 109.95} | |
| [INFO|2025-04-01 10:24:28] logging.py:143 >> {'loss': 0.9451, 'learning_rate': 2.9935e-05, 'epoch': 1.31, 'throughput': 109.94} | |
| [INFO|2025-04-01 10:25:22] logging.py:143 >> {'loss': 0.8202, 'learning_rate': 2.9548e-05, 'epoch': 1.32, 'throughput': 110.03} | |
| [INFO|2025-04-01 10:26:14] logging.py:143 >> {'loss': 0.9773, 'learning_rate': 2.9160e-05, 'epoch': 1.34, 'throughput': 110.05} | |
| [INFO|2025-04-01 10:27:04] logging.py:143 >> {'loss': 0.9101, 'learning_rate': 2.8771e-05, 'epoch': 1.35, 'throughput': 109.98} | |
| [INFO|2025-04-01 10:27:57] logging.py:143 >> {'loss': 0.9633, 'learning_rate': 2.8380e-05, 'epoch': 1.37, 'throughput': 110.01} | |
| [INFO|2025-04-01 10:28:47] logging.py:143 >> {'loss': 0.8886, 'learning_rate': 2.7989e-05, 'epoch': 1.38, 'throughput': 109.96} | |
| [INFO|2025-04-01 10:29:39] logging.py:143 >> {'loss': 0.9258, 'learning_rate': 2.7598e-05, 'epoch': 1.40, 'throughput': 109.96} | |
| [INFO|2025-04-01 10:30:30] logging.py:143 >> {'loss': 0.9039, 'learning_rate': 2.7205e-05, 'epoch': 1.41, 'throughput': 109.93} | |
| [INFO|2025-04-01 10:31:24] logging.py:143 >> {'loss': 1.0116, 'learning_rate': 2.6812e-05, 'epoch': 1.43, 'throughput': 109.99} | |
| [INFO|2025-04-01 10:32:14] logging.py:143 >> {'loss': 0.8218, 'learning_rate': 2.6419e-05, 'epoch': 1.44, 'throughput': 109.97} | |
| [INFO|2025-04-01 10:33:07] logging.py:143 >> {'loss': 0.8604, 'learning_rate': 2.6025e-05, 'epoch': 1.46, 'throughput': 109.96} | |
| [INFO|2025-04-01 10:33:59] logging.py:143 >> {'loss': 0.8044, 'learning_rate': 2.5631e-05, 'epoch': 1.47, 'throughput': 110.00} | |
| [INFO|2025-04-01 10:34:51] logging.py:143 >> {'loss': 0.9198, 'learning_rate': 2.5237e-05, 'epoch': 1.49, 'throughput': 109.98} | |
| [INFO|2025-04-01 10:35:45] logging.py:143 >> {'loss': 0.9181, 'learning_rate': 2.4842e-05, 'epoch': 1.50, 'throughput': 109.99} | |
| [INFO|2025-04-01 10:35:45] trainer.py:3966 >> Saving model checkpoint to saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-500 | |
| [INFO|2025-04-01 10:35:45] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 10:35:45] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 10:35:45] tokenization_utils_base.py:2510 >> tokenizer config file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-500/tokenizer_config.json | |
| [INFO|2025-04-01 10:35:45] tokenization_utils_base.py:2519 >> Special tokens file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-500/special_tokens_map.json | |
| [INFO|2025-04-01 10:36:39] logging.py:143 >> {'loss': 0.8644, 'learning_rate': 2.4448e-05, 'epoch': 1.52, 'throughput': 110.01} | |
| [INFO|2025-04-01 10:37:33] logging.py:143 >> {'loss': 1.0127, 'learning_rate': 2.4054e-05, 'epoch': 1.53, 'throughput': 110.05} | |
| [INFO|2025-04-01 10:38:24] logging.py:143 >> {'loss': 0.7937, 'learning_rate': 2.3660e-05, 'epoch': 1.55, 'throughput': 110.02} | |
| [INFO|2025-04-01 10:39:15] logging.py:143 >> {'loss': 0.9806, 'learning_rate': 2.3267e-05, 'epoch': 1.56, 'throughput': 110.01} | |
| [INFO|2025-04-01 10:40:09] logging.py:143 >> {'loss': 0.9340, 'learning_rate': 2.2873e-05, 'epoch': 1.58, 'throughput': 110.08} | |
| [INFO|2025-04-01 10:41:02] logging.py:143 >> {'loss': 0.9288, 'learning_rate': 2.2481e-05, 'epoch': 1.59, 'throughput': 110.09} | |
| [INFO|2025-04-01 10:41:55] logging.py:143 >> {'loss': 0.8597, 'learning_rate': 2.2089e-05, 'epoch': 1.61, 'throughput': 110.07} | |
| [INFO|2025-04-01 10:42:48] logging.py:143 >> {'loss': 0.8817, 'learning_rate': 2.1698e-05, 'epoch': 1.62, 'throughput': 110.10} | |
| [INFO|2025-04-01 10:43:41] logging.py:143 >> {'loss': 0.7770, 'learning_rate': 2.1307e-05, 'epoch': 1.64, 'throughput': 110.13} | |
| [INFO|2025-04-01 10:44:34] logging.py:143 >> {'loss': 0.7980, 'learning_rate': 2.0918e-05, 'epoch': 1.65, 'throughput': 110.11} | |
| [INFO|2025-04-01 10:45:28] logging.py:143 >> {'loss': 0.9104, 'learning_rate': 2.0529e-05, 'epoch': 1.67, 'throughput': 110.17} | |
| [INFO|2025-04-01 10:46:22] logging.py:143 >> {'loss': 0.8293, 'learning_rate': 2.0142e-05, 'epoch': 1.68, 'throughput': 110.26} | |
| [INFO|2025-04-01 10:47:14] logging.py:143 >> {'loss': 0.8821, 'learning_rate': 1.9756e-05, 'epoch': 1.70, 'throughput': 110.23} | |
| [INFO|2025-04-01 10:48:07] logging.py:143 >> {'loss': 0.8253, 'learning_rate': 1.9371e-05, 'epoch': 1.71, 'throughput': 110.27} | |
| [INFO|2025-04-01 10:48:59] logging.py:143 >> {'loss': 0.9391, 'learning_rate': 1.8988e-05, 'epoch': 1.73, 'throughput': 110.25} | |
| [INFO|2025-04-01 10:49:52] logging.py:143 >> {'loss': 0.8711, 'learning_rate': 1.8606e-05, 'epoch': 1.74, 'throughput': 110.28} | |
| [INFO|2025-04-01 10:50:44] logging.py:143 >> {'loss': 0.8346, 'learning_rate': 1.8225e-05, 'epoch': 1.76, 'throughput': 110.26} | |
| [INFO|2025-04-01 10:51:36] logging.py:143 >> {'loss': 0.8275, 'learning_rate': 1.7847e-05, 'epoch': 1.77, 'throughput': 110.27} | |
| [INFO|2025-04-01 10:52:27] logging.py:143 >> {'loss': 0.9435, 'learning_rate': 1.7470e-05, 'epoch': 1.79, 'throughput': 110.27} | |
| [INFO|2025-04-01 10:53:20] logging.py:143 >> {'loss': 0.8584, 'learning_rate': 1.7095e-05, 'epoch': 1.80, 'throughput': 110.29} | |
| [INFO|2025-04-01 10:53:20] trainer.py:3966 >> Saving model checkpoint to saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-600 | |
| [INFO|2025-04-01 10:53:20] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 10:53:20] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 10:53:21] tokenization_utils_base.py:2510 >> tokenizer config file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-600/tokenizer_config.json | |
| [INFO|2025-04-01 10:53:21] tokenization_utils_base.py:2519 >> Special tokens file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-600/special_tokens_map.json | |
| [INFO|2025-04-01 10:54:14] logging.py:143 >> {'loss': 0.8800, 'learning_rate': 1.6722e-05, 'epoch': 1.82, 'throughput': 110.30} | |
| [INFO|2025-04-01 10:55:06] logging.py:143 >> {'loss': 0.8825, 'learning_rate': 1.6351e-05, 'epoch': 1.83, 'throughput': 110.30} | |
| [INFO|2025-04-01 10:56:00] logging.py:143 >> {'loss': 0.9978, 'learning_rate': 1.5982e-05, 'epoch': 1.85, 'throughput': 110.33} | |
| [INFO|2025-04-01 10:56:54] logging.py:143 >> {'loss': 0.9626, 'learning_rate': 1.5615e-05, 'epoch': 1.86, 'throughput': 110.38} | |
| [INFO|2025-04-01 10:57:45] logging.py:143 >> {'loss': 0.9308, 'learning_rate': 1.5251e-05, 'epoch': 1.88, 'throughput': 110.37} | |
| [INFO|2025-04-01 10:58:39] logging.py:143 >> {'loss': 0.9757, 'learning_rate': 1.4889e-05, 'epoch': 1.89, 'throughput': 110.38} | |
| [INFO|2025-04-01 10:59:33] logging.py:143 >> {'loss': 0.7670, 'learning_rate': 1.4530e-05, 'epoch': 1.91, 'throughput': 110.43} | |
| [INFO|2025-04-01 11:00:23] logging.py:143 >> {'loss': 0.9272, 'learning_rate': 1.4173e-05, 'epoch': 1.92, 'throughput': 110.40} | |
| [INFO|2025-04-01 11:01:16] logging.py:143 >> {'loss': 0.7941, 'learning_rate': 1.3819e-05, 'epoch': 1.94, 'throughput': 110.41} | |
| [INFO|2025-04-01 11:02:08] logging.py:143 >> {'loss': 0.8408, 'learning_rate': 1.3468e-05, 'epoch': 1.95, 'throughput': 110.38} | |
| [INFO|2025-04-01 11:03:00] logging.py:143 >> {'loss': 0.8459, 'learning_rate': 1.3120e-05, 'epoch': 1.97, 'throughput': 110.35} | |
| [INFO|2025-04-01 11:03:52] logging.py:143 >> {'loss': 1.0117, 'learning_rate': 1.2774e-05, 'epoch': 1.98, 'throughput': 110.36} | |
| [INFO|2025-04-01 11:04:43] logging.py:143 >> {'loss': 0.9665, 'learning_rate': 1.2432e-05, 'epoch': 2.00, 'throughput': 110.32} | |
| [INFO|2025-04-01 11:05:28] logging.py:143 >> {'loss': 0.7625, 'learning_rate': 1.2093e-05, 'epoch': 2.01, 'throughput': 110.34} | |
| [INFO|2025-04-01 11:06:21] logging.py:143 >> {'loss': 0.8667, 'learning_rate': 1.1756e-05, 'epoch': 2.03, 'throughput': 110.33} | |
| [INFO|2025-04-01 11:07:14] logging.py:143 >> {'loss': 0.8297, 'learning_rate': 1.1424e-05, 'epoch': 2.04, 'throughput': 110.34} | |
| [INFO|2025-04-01 11:08:06] logging.py:143 >> {'loss': 0.8774, 'learning_rate': 1.1094e-05, 'epoch': 2.06, 'throughput': 110.35} | |
| [INFO|2025-04-01 11:08:59] logging.py:143 >> {'loss': 0.8476, 'learning_rate': 1.0768e-05, 'epoch': 2.07, 'throughput': 110.37} | |
| [INFO|2025-04-01 11:09:51] logging.py:143 >> {'loss': 0.8641, 'learning_rate': 1.0446e-05, 'epoch': 2.09, 'throughput': 110.35} | |
| [INFO|2025-04-01 11:10:44] logging.py:143 >> {'loss': 0.8383, 'learning_rate': 1.0127e-05, 'epoch': 2.10, 'throughput': 110.36} | |
| [INFO|2025-04-01 11:10:44] trainer.py:3966 >> Saving model checkpoint to saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-700 | |
| [INFO|2025-04-01 11:10:45] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 11:10:45] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 11:10:45] tokenization_utils_base.py:2510 >> tokenizer config file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-700/tokenizer_config.json | |
| [INFO|2025-04-01 11:10:45] tokenization_utils_base.py:2519 >> Special tokens file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-700/special_tokens_map.json | |
| [INFO|2025-04-01 11:11:38] logging.py:143 >> {'loss': 0.9123, 'learning_rate': 9.8123e-06, 'epoch': 2.12, 'throughput': 110.32} | |
| [INFO|2025-04-01 11:12:28] logging.py:143 >> {'loss': 0.9635, 'learning_rate': 9.5010e-06, 'epoch': 2.13, 'throughput': 110.28} | |
| [INFO|2025-04-01 11:13:21] logging.py:143 >> {'loss': 0.9221, 'learning_rate': 9.1936e-06, 'epoch': 2.15, 'throughput': 110.30} | |
| [INFO|2025-04-01 11:14:16] logging.py:143 >> {'loss': 0.8757, 'learning_rate': 8.8901e-06, 'epoch': 2.16, 'throughput': 110.34} | |
| [INFO|2025-04-01 11:15:08] logging.py:143 >> {'loss': 0.7958, 'learning_rate': 8.5906e-06, 'epoch': 2.18, 'throughput': 110.36} | |
| [INFO|2025-04-01 11:16:01] logging.py:143 >> {'loss': 0.7993, 'learning_rate': 8.2952e-06, 'epoch': 2.19, 'throughput': 110.36} | |
| [INFO|2025-04-01 11:16:55] logging.py:143 >> {'loss': 0.8436, 'learning_rate': 8.0039e-06, 'epoch': 2.21, 'throughput': 110.37} | |
| [INFO|2025-04-01 11:17:48] logging.py:143 >> {'loss': 0.8960, 'learning_rate': 7.7169e-06, 'epoch': 2.22, 'throughput': 110.38} | |
| [INFO|2025-04-01 11:18:39] logging.py:143 >> {'loss': 0.8948, 'learning_rate': 7.4342e-06, 'epoch': 2.24, 'throughput': 110.35} | |
| [INFO|2025-04-01 11:19:31] logging.py:143 >> {'loss': 0.8546, 'learning_rate': 7.1558e-06, 'epoch': 2.25, 'throughput': 110.33} | |
| [INFO|2025-04-01 11:20:23] logging.py:143 >> {'loss': 0.8494, 'learning_rate': 6.8819e-06, 'epoch': 2.27, 'throughput': 110.35} | |
| [INFO|2025-04-01 11:21:17] logging.py:143 >> {'loss': 0.7723, 'learning_rate': 6.6125e-06, 'epoch': 2.28, 'throughput': 110.40} | |
| [INFO|2025-04-01 11:22:08] logging.py:143 >> {'loss': 0.9168, 'learning_rate': 6.3477e-06, 'epoch': 2.30, 'throughput': 110.37} | |
| [INFO|2025-04-01 11:23:01] logging.py:143 >> {'loss': 0.8831, 'learning_rate': 6.0875e-06, 'epoch': 2.31, 'throughput': 110.38} | |
| [INFO|2025-04-01 11:23:52] logging.py:143 >> {'loss': 0.8540, 'learning_rate': 5.8320e-06, 'epoch': 2.33, 'throughput': 110.32} | |
| [INFO|2025-04-01 11:24:43] logging.py:143 >> {'loss': 0.8843, 'learning_rate': 5.5813e-06, 'epoch': 2.34, 'throughput': 110.32} | |
| [INFO|2025-04-01 11:25:36] logging.py:143 >> {'loss': 0.8620, 'learning_rate': 5.3354e-06, 'epoch': 2.36, 'throughput': 110.34} | |
| [INFO|2025-04-01 11:26:29] logging.py:143 >> {'loss': 0.9850, 'learning_rate': 5.0944e-06, 'epoch': 2.37, 'throughput': 110.34} | |
| [INFO|2025-04-01 11:27:20] logging.py:143 >> {'loss': 0.7679, 'learning_rate': 4.8583e-06, 'epoch': 2.39, 'throughput': 110.31} | |
| [INFO|2025-04-01 11:28:11] logging.py:143 >> {'loss': 0.8198, 'learning_rate': 4.6273e-06, 'epoch': 2.40, 'throughput': 110.30} | |
| [INFO|2025-04-01 11:28:11] trainer.py:3966 >> Saving model checkpoint to saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-800 | |
| [INFO|2025-04-01 11:28:12] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 11:28:12] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 11:28:12] tokenization_utils_base.py:2510 >> tokenizer config file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-800/tokenizer_config.json | |
| [INFO|2025-04-01 11:28:12] tokenization_utils_base.py:2519 >> Special tokens file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-800/special_tokens_map.json | |
| [INFO|2025-04-01 11:29:06] logging.py:143 >> {'loss': 0.7773, 'learning_rate': 4.4013e-06, 'epoch': 2.42, 'throughput': 110.28} | |
| [INFO|2025-04-01 11:29:58] logging.py:143 >> {'loss': 0.9312, 'learning_rate': 4.1805e-06, 'epoch': 2.43, 'throughput': 110.27} | |
| [INFO|2025-04-01 11:30:50] logging.py:143 >> {'loss': 0.8497, 'learning_rate': 3.9648e-06, 'epoch': 2.45, 'throughput': 110.28} | |
| [INFO|2025-04-01 11:31:43] logging.py:143 >> {'loss': 0.7820, 'learning_rate': 3.7543e-06, 'epoch': 2.46, 'throughput': 110.29} | |
| [INFO|2025-04-01 11:32:36] logging.py:143 >> {'loss': 0.8937, 'learning_rate': 3.5492e-06, 'epoch': 2.48, 'throughput': 110.31} | |
| [INFO|2025-04-01 11:33:29] logging.py:143 >> {'loss': 0.7039, 'learning_rate': 3.3494e-06, 'epoch': 2.49, 'throughput': 110.36} | |
| [INFO|2025-04-01 11:34:22] logging.py:143 >> {'loss': 0.9265, 'learning_rate': 3.1549e-06, 'epoch': 2.51, 'throughput': 110.36} | |
| [INFO|2025-04-01 11:35:13] logging.py:143 >> {'loss': 0.8669, 'learning_rate': 2.9659e-06, 'epoch': 2.52, 'throughput': 110.36} | |
| [INFO|2025-04-01 11:36:05] logging.py:143 >> {'loss': 0.9174, 'learning_rate': 2.7824e-06, 'epoch': 2.54, 'throughput': 110.37} | |
| [INFO|2025-04-01 11:36:58] logging.py:143 >> {'loss': 0.8718, 'learning_rate': 2.6044e-06, 'epoch': 2.55, 'throughput': 110.38} | |
| [INFO|2025-04-01 11:37:50] logging.py:143 >> {'loss': 0.8634, 'learning_rate': 2.4320e-06, 'epoch': 2.57, 'throughput': 110.37} | |
| [INFO|2025-04-01 11:38:43] logging.py:143 >> {'loss': 0.8450, 'learning_rate': 2.2652e-06, 'epoch': 2.58, 'throughput': 110.38} | |
| [INFO|2025-04-01 11:39:34] logging.py:143 >> {'loss': 0.8008, 'learning_rate': 2.1040e-06, 'epoch': 2.60, 'throughput': 110.36} | |
| [INFO|2025-04-01 11:40:26] logging.py:143 >> {'loss': 0.8797, 'learning_rate': 1.9485e-06, 'epoch': 2.61, 'throughput': 110.36} | |
| [INFO|2025-04-01 11:41:19] logging.py:143 >> {'loss': 0.9460, 'learning_rate': 1.7988e-06, 'epoch': 2.63, 'throughput': 110.37} | |
| [INFO|2025-04-01 11:42:10] logging.py:143 >> {'loss': 0.8032, 'learning_rate': 1.6548e-06, 'epoch': 2.64, 'throughput': 110.36} | |
| [INFO|2025-04-01 11:43:02] logging.py:143 >> {'loss': 0.8892, 'learning_rate': 1.5167e-06, 'epoch': 2.66, 'throughput': 110.37} | |
| [INFO|2025-04-01 11:43:56] logging.py:143 >> {'loss': 0.8560, 'learning_rate': 1.3844e-06, 'epoch': 2.67, 'throughput': 110.39} | |
| [INFO|2025-04-01 11:44:49] logging.py:143 >> {'loss': 0.8617, 'learning_rate': 1.2579e-06, 'epoch': 2.69, 'throughput': 110.43} | |
| [INFO|2025-04-01 11:45:43] logging.py:143 >> {'loss': 0.9117, 'learning_rate': 1.1374e-06, 'epoch': 2.70, 'throughput': 110.46} | |
| [INFO|2025-04-01 11:45:43] trainer.py:3966 >> Saving model checkpoint to saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-900 | |
| [INFO|2025-04-01 11:45:44] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 11:45:44] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 11:45:44] tokenization_utils_base.py:2510 >> tokenizer config file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-900/tokenizer_config.json | |
| [INFO|2025-04-01 11:45:44] tokenization_utils_base.py:2519 >> Special tokens file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-900/special_tokens_map.json | |
| [INFO|2025-04-01 11:46:37] logging.py:143 >> {'loss': 0.7855, 'learning_rate': 1.0228e-06, 'epoch': 2.72, 'throughput': 110.44} | |
| [INFO|2025-04-01 11:47:29] logging.py:143 >> {'loss': 0.8212, 'learning_rate': 9.1416e-07, 'epoch': 2.73, 'throughput': 110.42} | |
| [INFO|2025-04-01 11:48:22] logging.py:143 >> {'loss': 0.8404, 'learning_rate': 8.1152e-07, 'epoch': 2.75, 'throughput': 110.45} | |
| [INFO|2025-04-01 11:49:15] logging.py:143 >> {'loss': 0.7782, 'learning_rate': 7.1489e-07, 'epoch': 2.76, 'throughput': 110.45} | |
| [INFO|2025-04-01 11:50:08] logging.py:143 >> {'loss': 0.7847, 'learning_rate': 6.2430e-07, 'epoch': 2.78, 'throughput': 110.45} | |
| [INFO|2025-04-01 11:51:01] logging.py:143 >> {'loss': 0.8857, 'learning_rate': 5.3977e-07, 'epoch': 2.79, 'throughput': 110.46} | |
| [INFO|2025-04-01 11:51:54] logging.py:143 >> {'loss': 0.8029, 'learning_rate': 4.6133e-07, 'epoch': 2.81, 'throughput': 110.48} | |
| [INFO|2025-04-01 11:52:46] logging.py:143 >> {'loss': 0.8154, 'learning_rate': 3.8899e-07, 'epoch': 2.82, 'throughput': 110.45} | |
| [INFO|2025-04-01 11:53:39] logging.py:143 >> {'loss': 0.8791, 'learning_rate': 3.2277e-07, 'epoch': 2.84, 'throughput': 110.45} | |
| [INFO|2025-04-01 11:54:32] logging.py:143 >> {'loss': 0.7870, 'learning_rate': 2.6269e-07, 'epoch': 2.85, 'throughput': 110.46} | |
| [INFO|2025-04-01 11:55:26] logging.py:143 >> {'loss': 0.8831, 'learning_rate': 2.0876e-07, 'epoch': 2.87, 'throughput': 110.49} | |
| [INFO|2025-04-01 11:56:19] logging.py:143 >> {'loss': 0.7677, 'learning_rate': 1.6100e-07, 'epoch': 2.88, 'throughput': 110.49} | |
| [INFO|2025-04-01 11:57:10] logging.py:143 >> {'loss': 0.7567, 'learning_rate': 1.1942e-07, 'epoch': 2.90, 'throughput': 110.47} | |
| [INFO|2025-04-01 11:58:02] logging.py:143 >> {'loss': 0.8944, 'learning_rate': 8.4022e-08, 'epoch': 2.91, 'throughput': 110.46} | |
| [INFO|2025-04-01 11:58:53] logging.py:143 >> {'loss': 0.9737, 'learning_rate': 5.4824e-08, 'epoch': 2.93, 'throughput': 110.44} | |
| [INFO|2025-04-01 11:59:45] logging.py:143 >> {'loss': 0.8965, 'learning_rate': 3.1830e-08, 'epoch': 2.95, 'throughput': 110.44} | |
| [INFO|2025-04-01 12:00:37] logging.py:143 >> {'loss': 0.8370, 'learning_rate': 1.5046e-08, 'epoch': 2.96, 'throughput': 110.41} | |
| [INFO|2025-04-01 12:01:29] logging.py:143 >> {'loss': 0.7812, 'learning_rate': 4.4769e-09, 'epoch': 2.98, 'throughput': 110.40} | |
| [INFO|2025-04-01 12:02:24] logging.py:143 >> {'loss': 0.8613, 'learning_rate': 1.2436e-10, 'epoch': 2.99, 'throughput': 110.45} | |
| [INFO|2025-04-01 12:02:35] trainer.py:3966 >> Saving model checkpoint to saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-996 | |
| [INFO|2025-04-01 12:02:35] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 12:02:35] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 12:02:35] tokenization_utils_base.py:2510 >> tokenizer config file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-996/tokenizer_config.json | |
| [INFO|2025-04-01 12:02:35] tokenization_utils_base.py:2519 >> Special tokens file saved in saves/Custom/lora/train_2025-04-01-09-06-36/checkpoint-996/special_tokens_map.json | |
| [INFO|2025-04-01 12:02:36] trainer.py:2665 >> | |
| Training completed. Do not forget to share your model on huggingface.co/models =) | |
| [INFO|2025-04-01 12:02:36] trainer.py:3966 >> Saving model checkpoint to saves/Custom/lora/train_2025-04-01-09-06-36 | |
| [INFO|2025-04-01 12:02:37] configuration_utils.py:699 >> loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--prithivMLmods--Qwen2-VL-OCR-2B-Instruct/snapshots/a54254d5cc9f82e1c362db82adede275d20bbc6b/config.json | |
| [INFO|2025-04-01 12:02:37] configuration_utils.py:771 >> Model config Qwen2VLConfig { | |
| "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, | |
| "pad_token_id": 151654, | |
| "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": "bfloat16", | |
| "transformers_version": "4.50.0", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "video_token_id": 151656, | |
| "vision_config": { | |
| "depth": 32, | |
| "embed_dim": 1280, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1536, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "mlp_ratio": 4, | |
| "model_type": "qwen2_vl", | |
| "num_heads": 16, | |
| "patch_size": 14, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "temporal_patch_size": 2 | |
| }, | |
| "vision_end_token_id": 151653, | |
| "vision_start_token_id": 151652, | |
| "vision_token_id": 151654, | |
| "vocab_size": 151936 | |
| } | |
| [INFO|2025-04-01 12:02:37] tokenization_utils_base.py:2510 >> tokenizer config file saved in saves/Custom/lora/train_2025-04-01-09-06-36/tokenizer_config.json | |
| [INFO|2025-04-01 12:02:37] tokenization_utils_base.py:2519 >> Special tokens file saved in saves/Custom/lora/train_2025-04-01-09-06-36/special_tokens_map.json | |
| [WARNING|2025-04-01 12:02:37] logging.py:148 >> No metric eval_loss to plot. | |
| [WARNING|2025-04-01 12:02:37] logging.py:148 >> No metric eval_accuracy to plot. | |
| [INFO|2025-04-01 12:02:37] modelcard.py:449 >> Dropping the following result as it does not have all the necessary fields: | |
| {'task': {'name': 'Causal Language Modeling', 'type': 'text-generation'}} | |