Training in progress, epoch 1

README.md

@@ -1,5 +1,5 @@
 ---
-base_model: Qwen/Qwen2.5-VL-7B-Instruct
+base_model: UCSC-VLAA/VLAA-Thinker-Qwen2.5VL-3B
 library_name: transformers
 model_name: qwen2.5-VL-7B-instruct-00004-VqCaAuuoeWk_0
 tags:
@@ -11,7 +11,7 @@ licence: license
 
 # Model Card for qwen2.5-VL-7B-instruct-00004-VqCaAuuoeWk_0
 
-This model is a fine-tuned version of [Qwen/Qwen2.5-VL-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct).
+This model is a fine-tuned version of [UCSC-VLAA/VLAA-Thinker-Qwen2.5VL-3B](https://huggingface.co/UCSC-VLAA/VLAA-Thinker-Qwen2.5VL-3B).
 It has been trained using [TRL](https://github.com/huggingface/trl).
 
 ## Quick start

adapter_config.json

@@ -1,7 +1,7 @@
 {
   "alpha_pattern": {},
   "auto_mapping": null,
-  "base_model_name_or_path": "Qwen/Qwen2.5-VL-7B-Instruct",
+  "base_model_name_or_path": "UCSC-VLAA/VLAA-Thinker-Qwen2.5VL-3B",
   "bias": "none",
   "corda_config": null,
   "eva_config": null,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:832421a1bed0ec02e349d8746f3b9b345e38ead2c05a3641b850acc29b629963
-size 10108960
+oid sha256:60c39a0b2982363789261b33e20a51e2158f3f232d8ab89530a449a4a0b5e67a
+size 7393888

preprocessor_config.json

@@ -21,7 +21,7 @@
     0.27577711
   ],
   "input_data_format": null,
-  "max_pixels": 12845056,
+  "max_pixels": 501760,
   "merge_size": 2,
   "min_pixels": 3136,
   "patch_size": 14,
@@ -30,7 +30,7 @@
   "rescale_factor": 0.00392156862745098,
   "return_tensors": null,
   "size": {
-    "longest_edge": 12845056,
+    "longest_edge": 501760,
     "shortest_edge": 3136
   },
   "temporal_patch_size": 2

runs/Sep09_15-32-32_workstation-one/events.out.tfevents.1757431963.workstation-one.1228196.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:252cfbd48232339693b25f34e04b45cd539ce56267d4ee69b1452246dd2752f0
+size 10747

tokenizer_config.json

@@ -199,10 +199,13 @@
   "eos_token": "<|im_end|>",
   "errors": "replace",
   "extra_special_tokens": {},
+  "max_pixels": 501760,
+  "min_pixels": 3136,
   "model_max_length": 131072,
   "pad_token": "<|endoftext|>",
   "processor_class": "Qwen2_5_VLProcessor",
   "split_special_tokens": false,
   "tokenizer_class": "Qwen2Tokenizer",
-  "unk_token": null
+  "unk_token": null,
+  "use_fast": true
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3dd1341587efc15903605ae5e5e373891d4c6b006a9ae24f855c03d1e1373dcc
-size 6353
+oid sha256:6f4a607f106199b8cada31e14d82dc5dffda3c28c9eca97ec41c741f684d2366
+size 6417

video_preprocessor_config.json

@@ -16,6 +16,7 @@
     0.4578275,
     0.40821073
   ],
+  "image_processor_type": "Qwen2VLImageProcessorFast",
   "image_std": [
     0.26862954,
     0.26130258,
@@ -23,7 +24,7 @@
   ],
   "input_data_format": null,
   "max_frames": 768,
-  "max_pixels": 12845056,
+  "max_pixels": 501760,
   "merge_size": 2,
   "min_frames": 4,
   "min_pixels": 3136,
@@ -34,7 +35,7 @@
   "rescale_factor": 0.00392156862745098,
   "return_metadata": false,
   "size": {
-    "longest_edge": 12845056,
+    "longest_edge": 501760,
     "shortest_edge": 3136
   },
   "size_divisor": null,