Upload qwen3vl_sft_v16 excluding caches

.gitattributes

@@ -55,3 +55,4 @@ sft/qwen3vl_sft_v7_vit/evaluate_best/evaluation_detail.json filter=lfs diff=lfs
 sft/qwen3vl_sft_v8_multitask/evaluate_best/evaluation_detail.json filter=lfs diff=lfs merge=lfs -text
 sft/qwen3vl_sft_v9_multitask_llava/evaluate_best/evaluation_detail.json filter=lfs diff=lfs merge=lfs -text
 sft/qwen3vl_sft_v18/evaluate_best/evaluation_detail.json filter=lfs diff=lfs merge=lfs -text
+sft/qwen3vl_sft_v16/evaluate_best/evaluation_detail.json filter=lfs diff=lfs merge=lfs -text

sft/qwen3vl_sft_v16/evaluate_best/evaluation.text

@@ -0,0 +1,199 @@
+书法 bbox 召回率与参考文本相似度评测（0–5）
+judge：gpt-6-luna
+已保存样本：2188/2188；状态：{"ok":2188}
+中断：False；全局 API 错误：None
+局部分数仅评 IoU 命中的文本；主均值先逐图平均再跨图平均，只有该字段完成评分的图像计入。
+分数表示与参考的语义、内容覆盖及专业概念一致性，不是图像事实准确性。
+参考字段缺失则跳过；预测字段为空得 0；API 失败不当作 0。
+严格联合成功要求命中 bbox 且每个必要 LLM judge 字段均 ≥ 4；ROUGE-L/BERTScore 不参与合格判定。
+整体评论与 IoU 阈值无关，每字评一次。
+
+========== IoU ≥ 0.3 ==========
+笔法
+参考 4588，命中 2654；无效参考框 0
+召回率：57.85%；逐图宏召回率：62.14%；匹配平均 IoU：0.7935
+严格联合成功：171 条；联合成功率：3.73%；命中后合格率：6.44%；未合格命中 2483，未完成评分命中 0
+  visual_evidence：逐图均值 2.9319/5；区域均值 2.9461/5；已评 2654/2654，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.3725；区域均值 0.3732；已评 2654，缺失 0
+    bertscore_precision：逐图均值 0.7761；区域均值 0.7761；已评 2654，缺失 0
+    bertscore_recall：逐图均值 0.7682；区域均值 0.7685；已评 2654，缺失 0
+    bertscore_f1：逐图均值 0.7719；区域均值 0.7720；已评 2654，缺失 0
+  technique：逐图均值 1.8739/5；区域均值 1.8787/5；已评 2654/2654，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.7326；区域均值 0.7301；已评 2654，缺失 0
+    bertscore_precision：逐图均值 0.8536；区域均值 0.8525；已评 2654，缺失 0
+    bertscore_recall：逐图均值 0.8417；区域均值 0.8407；已评 2654，缺失 0
+    bertscore_f1：逐图均值 0.8471；区域均值 0.8461；已评 2654，缺失 0
+  technical_explanation：逐图均值 2.6203/5；区域均值 2.6089/5；已评 2654/2654，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.3434；区域均值 0.3426；已评 2654，缺失 0
+    bertscore_precision：逐图均值 0.7676；区域均值 0.7665；已评 2654，缺失 0
+    bertscore_recall：逐图均值 0.7579；区域均值 0.7574；已评 2654，缺失 0
+    bertscore_f1：逐图均值 0.7625；区域均值 0.7617；已评 2654，缺失 0
+  算法指标（该类型全部局部文本字段合计）：
+    rouge_l_f1：区域均值 0.4819；字段-图像均值 0.4828；已评 7962，缺失 0
+    bertscore_precision：区域均值 0.7984；字段-图像均值 0.7991；已评 7962，缺失 0
+    bertscore_recall：区域均值 0.7889；字段-图像均值 0.7893；已评 7962，缺失 0
+    bertscore_f1：区域均值 0.7933；字段-图像均值 0.7938；已评 7962，缺失 0
+
+字法
+参考 4540，命中 1744；无效参考框 0
+召回率：38.41%；逐图宏召回率：39.22%；匹配平均 IoU：0.7413
+严格联合成功：73 条；联合成功率：1.61%；命中后合格率：4.19%；未合格命中 1671，未完成评分命中 0
+  evidence：逐图均值 2.0406/5；区域均值 2.0080/5；已评 1744/1744，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.2962；区域均值 0.2902；已评 1744，缺失 0
+    bertscore_precision：逐图均值 0.7349；区域均值 0.7322；已评 1744，缺失 0
+    bertscore_recall：逐图均值 0.7302；区域均值 0.7270；已评 1744，缺失 0
+    bertscore_f1：逐图均值 0.7322；区域均值 0.7293；已评 1744，缺失 0
+  description：逐图均值 2.0066/5；区域均值 1.9685/5；已评 1744/1744，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.3027；区域均值 0.2932；已评 1744，缺失 0
+    bertscore_precision：逐图均值 0.7130；区域均值 0.7072；已评 1744，缺失 0
+    bertscore_recall：逐图均值 0.7129；区域均值 0.7066；已评 1744，缺失 0
+    bertscore_f1：逐图均值 0.7122；区域均值 0.7061；已评 1744，缺失 0
+  feature：逐图均值 1.6767/5；区域均值 1.6766/5；已评 1744/1744，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.1682；区域均值 0.1702；已评 1744，缺失 0
+    bertscore_precision：逐图均值 0.6341；区域均值 0.6337；已评 1744，缺失 0
+    bertscore_recall：逐图均值 0.6269；区域均值 0.6240；已评 1744，缺失 0
+    bertscore_f1：逐图均值 0.6298；区域均值 0.6280；已评 1744，缺失 0
+  effect：逐图均值 2.1909/5；区域均值 2.1525/5；已评 1744/1744，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.3098；区域均值 0.2989；已评 1744，缺失 0
+    bertscore_precision：逐图均值 0.7320；区域均值 0.7254；已评 1744，缺失 0
+    bertscore_recall：逐图均值 0.7311；区域均值 0.7250；已评 1744，缺失 0
+    bertscore_f1：逐图均值 0.7311；区域均值 0.7247；已评 1744，缺失 0
+  算法指标（该类型全部局部文本字段合计）：
+    rouge_l_f1：区域均值 0.2631；字段-图像均值 0.2692；已评 6976，缺失 0
+    bertscore_precision：区域均值 0.6996；字段-图像均值 0.7035；已评 6976，缺失 0
+    bertscore_recall：区域均值 0.6957；字段-图像均值 0.7003；已评 6976，缺失 0
+    bertscore_f1：区域均值 0.6970；字段-图像均值 0.7013；已评 6976，缺失 0
+
+算法指标（笔法+字法全部局部字段合计）：
+  rouge_l_f1：区域均值 0.3797；已评 14938，缺失 0
+  bertscore_precision：区域均值 0.7523；已评 14938，缺失 0
+  bertscore_recall：区域均值 0.7453；已评 14938，缺失 0
+  bertscore_f1：区域均值 0.7483；已评 14938，缺失 0
+========== IoU ≥ 0.5 ==========
+笔法
+参考 4588，命中 2258；无效参考框 0
+召回率：49.22%；逐图宏召回率：52.45%；匹配平均 IoU：0.8641
+严格联合成功：155 条；联合成功率：3.38%；命中后合格率：6.86%；未合格命中 2103，未完成评分命中 0
+  visual_evidence：逐图均值 3.0624/5；区域均值 3.0793/5；已评 2258/2258，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.3863；区域均值 0.3880；已评 2258，缺失 0
+    bertscore_precision：逐图均值 0.7810；区域均值 0.7818；已评 2258，缺失 0
+    bertscore_recall：逐图均值 0.7735；区域均值 0.7744；已评 2258，缺失 0
+    bertscore_f1：逐图均值 0.7770；区域均值 0.7779；已评 2258，缺失 0
+  technique：逐图均值 1.9439/5；区域均值 1.9411/5；已评 2258/2258，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.7384；区域均值 0.7367；已评 2258，缺失 0
+    bertscore_precision：逐图均值 0.8563；区域均值 0.8554；已评 2258，缺失 0
+    bertscore_recall：逐图均值 0.8450；区域均值 0.8443；已评 2258，缺失 0
+    bertscore_f1：逐图均值 0.8502；区域均值 0.8494；已评 2258，缺失 0
+  technical_explanation：逐图均值 2.7337/5；区域均值 2.7245/5；已评 2258/2258，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.3516；区域均值 0.3520；已评 2258，缺失 0
+    bertscore_precision：逐图均值 0.7713；区域均值 0.7710；已评 2258，缺失 0
+    bertscore_recall：逐图均值 0.7618；区域均值 0.7619；已评 2258，缺失 0
+    bertscore_f1：逐图均值 0.7663；区域均值 0.7662；已评 2258，缺失 0
+  算法指标（该类型全部局部文本字段合计）：
+    rouge_l_f1：区域均值 0.4922；字段-图像均值 0.4921；已评 6774，缺失 0
+    bertscore_precision：区域均值 0.8027；字段-图像均值 0.8029；已评 6774，缺失 0
+    bertscore_recall：区域均值 0.7935；字段-图像均值 0.7934；已评 6774，缺失 0
+    bertscore_f1：区域均值 0.7978；字段-图像均值 0.7978；已评 6774，缺失 0
+
+字法
+参考 4540，命中 1308；无效参考框 0
+召回率：28.81%；逐图宏召回率：27.94%；匹配平均 IoU：0.8584
+严格联合成功：71 条；联合成功率：1.56%；命中后合格率：5.43%；未合格命中 1237，未完成评分命中 0
+  evidence：逐图均值 2.1611/5；区域均值 2.1277/5；已评 1308/1308，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.3076；区域均值 0.3009；已评 1308，缺失 0
+    bertscore_precision：逐图均值 0.7390；区域均值 0.7362；已评 1308，缺失 0
+    bertscore_recall：逐图均值 0.7336；区域均值 0.7309；已评 1308，缺失 0
+    bertscore_f1：逐图均值 0.7359；区域均值 0.7332；已评 1308，缺失 0
+  description：逐图均值 2.1852/5；区域均值 2.1346/5；已评 1308/1308，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.3151；区域均值 0.3053；已评 1308，缺失 0
+    bertscore_precision：逐图均值 0.7171；区域均值 0.7112；已评 1308，缺失 0
+    bertscore_recall：逐图均值 0.7177；区域均值 0.7112；已评 1308，缺失 0
+    bertscore_f1：逐图均值 0.7166；区域均值 0.7104；已评 1308，缺失 0
+  feature：逐图均值 1.7806/5；区域均值 1.7844/5；已评 1308/1308，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.1928；区域均值 0.1977；已评 1308，缺失 0
+    bertscore_precision：逐图均值 0.6442；区域均值 0.6456；已评 1308，缺失 0
+    bertscore_recall：逐图均值 0.6362；区域均值 0.6353；已评 1308，缺失 0
+    bertscore_f1：逐图均值 0.6395；区域均值 0.6396；已评 1308，缺失 0
+  effect：逐图均值 2.2444/5；区域均值 2.2118/5；已评 1308/1308，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.3220；区域均值 0.3102；已评 1308，缺失 0
+    bertscore_precision：逐图均值 0.7361；区域均值 0.7288；已评 1308，缺失 0
+    bertscore_recall：逐图均值 0.7354；区域均值 0.7290；已评 1308，缺失 0
+    bertscore_f1：逐图均值 0.7353；区域均值 0.7284；已评 1308，缺失 0
+  算法指标（该类型全部局部文本字段合计）：
+    rouge_l_f1：区域均值 0.2785；字段-图像均值 0.2844；已评 5232，缺失 0
+    bertscore_precision：区域均值 0.7055；字段-图像均值 0.7091；已评 5232，缺失 0
+    bertscore_recall：区域均值 0.7016；字段-图像均值 0.7057；已评 5232，缺失 0
+    bertscore_f1：区域均值 0.7029；字段-图像均值 0.7068；已评 5232，缺失 0
+
+算法指标（笔法+字法全部局部字段合计）：
+  rouge_l_f1：区域均值 0.3991；已评 12006，缺失 0
+  bertscore_precision：区域均值 0.7603；已评 12006，缺失 0
+  bertscore_recall：区域均值 0.7535；已评 12006，缺失 0
+  bertscore_f1：区域均值 0.7565；已评 12006，缺失 0
+========== IoU ≥ 0.75 ==========
+笔法
+参考 4588，命中 1792；无效参考框 0
+召回率：39.06%；逐图宏召回率：41.37%；匹配平均 IoU：0.9263
+严格联合成功：129 条；联合成功率：2.81%；命中后合格率：7.20%；未合格命中 1663，未完成评分命中 0
+  visual_evidence：逐图均值 3.1556/5；区域均值 3.1641/5；已评 1792/1792，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.3974；区域均值 0.3990；已评 1792，缺失 0
+    bertscore_precision：逐图均值 0.7855；区域均值 0.7863；已评 1792，缺失 0
+    bertscore_recall：逐图均值 0.7783；区域均值 0.7791；已评 1792，缺失 0
+    bertscore_f1：逐图均值 0.7817；区域均值 0.7824；已评 1792，缺失 0
+  technique：逐图均值 1.9697/5；区域均值 1.9766/5；已评 1792/1792，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.7420；区域均值 0.7413；已评 1792，缺失 0
+    bertscore_precision：逐图均值 0.8583；区域均值 0.8580；已评 1792，缺失 0
+    bertscore_recall：逐图均值 0.8471；区域均值 0.8470；已评 1792，缺失 0
+    bertscore_f1：逐图均值 0.8522；区域均值 0.8520；已评 1792，缺失 0
+  technical_explanation：逐图均值 2.7955/5；区域均值 2.7896/5；已评 1792/1792，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.3582；区域均值 0.3591；已评 1792，缺失 0
+    bertscore_precision：逐图均值 0.7740；区域均值 0.7741；已评 1792，缺失 0
+    bertscore_recall：逐图均值 0.7645；区域均值 0.7648；已评 1792，缺失 0
+    bertscore_f1：逐图均值 0.7690；区域均值 0.7692；已评 1792，缺失 0
+  算法指标（该类型全部局部文本字段合计）：
+    rouge_l_f1：区域均值 0.4998；字段-图像均值 0.4992；已评 5376，缺失 0
+    bertscore_precision：区域均值 0.8061；字段-图像均值 0.8059；已评 5376，缺失 0
+    bertscore_recall：区域均值 0.7970；字段-图像均值 0.7966；已评 5376，缺失 0
+    bertscore_f1：区域均值 0.8012；字段-图像均值 0.8010；已评 5376，缺失 0
+
+字法
+参考 4540，命中 947；无效参考框 0
+召回率：20.86%；逐图宏召回率：19.35%；匹配平均 IoU：0.9533
+严格联合成功：56 条；联合成功率：1.23%；命中后合格率：5.91%；未合格命中 891，未完成评分命中 0
+  evidence：逐图均值 2.2222/5；区域均值 2.1911/5；已评 947/947，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.3102；区域均值 0.3032；已评 947，缺失 0
+    bertscore_precision：逐图均值 0.7404；区域均值 0.7372；已评 947，缺失 0
+    bertscore_recall：逐图均值 0.7354；区域均值 0.7321；已评 947，缺失 0
+    bertscore_f1：逐图均值 0.7375；区域均值 0.7343；已评 947，缺失 0
+  description：逐图均值 2.2725/5；区域均值 2.2302/5；已评 947/947，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.3223；区域均值 0.3122；已评 947，缺失 0
+    bertscore_precision：逐图均值 0.7169；区域均值 0.7111；已评 947，缺失 0
+    bertscore_recall：逐图均值 0.7195；区域均值 0.7135；已评 947，缺失 0
+    bertscore_f1：逐图均值 0.7173；区域均值 0.7114；已评 947，缺失 0
+  feature：逐图均值 1.8682/5；区域均值 1.8522/5；已评 947/947，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.2272；区域均值 0.2298；已评 947，缺失 0
+    bertscore_precision：逐图均值 0.6580；区域均值 0.6590；已评 947，缺失 0
+    bertscore_recall：逐图均值 0.6491；区域均值 0.6485；已评 947，缺失 0
+    bertscore_f1：逐图均值 0.6528；区域均值 0.6529；已评 947，缺失 0
+  effect：逐图均值 2.3035/5；区域均值 2.2661/5；已评 947/947，缺失评审 0，无参考跳过 0
+    rouge_l_f1：逐图均值 0.3280；区域均值 0.3170；已评 947，缺失 0
+    bertscore_precision：逐图均值 0.7356；区域均值 0.7287；已评 947，缺失 0
+    bertscore_recall：逐图均值 0.7367；区域均值 0.7308；已评 947，缺失 0
+    bertscore_f1：逐图均值 0.7356；区域均值 0.7292；已评 947，缺失 0
+  算法指标（该类型全部局部文本字段合计）：
+    rouge_l_f1：区���均值 0.2905；字段-图像均值 0.2969；已评 3788，缺失 0
+    bertscore_precision：区域均值 0.7090；字段-图像均值 0.7127；已评 3788，缺失 0
+    bertscore_recall：区域均值 0.7062；字段-图像均值 0.7102；已评 3788，缺失 0
+    bertscore_f1：区域均值 0.7070；字段-图像均值 0.7108；已评 3788，缺失 0
+
+算法指标（笔法+字法全部局部字段合计）：
+  rouge_l_f1：区域均值 0.4133；已评 9164，缺失 0
+  bertscore_precision：区域均值 0.7660；已评 9164，缺失 0
+  bertscore_recall：区域均值 0.7594；已评 9164，缺失 0
+  bertscore_f1：区域均值 0.7623；已评 9164，缺失 0
+========== 整体评论（所有阈值共享） ==========
+overall_comment：2.6307/5；已评 2188，无参考跳过 0，未评 0
+  rouge_l_f1：均值 0.2560；已评 2188，缺失 0
+  bertscore_precision：均值 0.7496；已评 2188，缺失 0
+  bertscore_recall：均值 0.7402；已评 2188，缺失 0
+  bertscore_f1：均值 0.7448；已评 2188，缺失 0

sft/qwen3vl_sft_v16/evaluate_best/evaluation_detail.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2f3700822364215221f62571c6fe7c3a9287f098af290db9152d60c74909c0b
+size 42794126

sft/qwen3vl_sft_v16/evaluate_best/evaluation_summary.json

@@ -0,0 +1,1104 @@
+{
+  "sample_count": 2188,
+  "requested_samples": 2188,
+  "interrupted": false,
+  "fatal_api_error": null,
+  "status_counts": {
+    "ok": 2188
+  },
+  "thresholds": {
+    "0.3": {
+      "bifa": {
+        "reference_count": 4588,
+        "hit_count": 2654,
+        "invalid_reference_bbox_count": 0,
+        "recall_micro": 0.5784655623365301,
+        "recall_image_macro": 0.6214461245266932,
+        "matched_iou_mean": 0.7935415741517535,
+        "qualification_rule": "all_required_fields",
+        "qualification_minimum_score": 4,
+        "qualified_analysis_count": 171,
+        "unqualified_hit_count": 2483,
+        "unscored_hit_count": 0,
+        "joint_success_rate": 0.03727114210985179,
+        "qualified_analysis_rate_on_bbox_hits": 0.06443104747550867,
+        "scores": {
+          "visual_evidence": {
+            "image_macro_mean": 2.9318743667679836,
+            "region_micro_mean": 2.9461190655614167,
+            "scored_regions": 2654,
+            "expected_matched_regions": 2654,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 1645,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.3725120764604666,
+                "region_micro_mean": 0.3731893506926202,
+                "scored_regions": 2654,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.7760778162840531,
+                "region_micro_mean": 0.7760641787990018,
+                "scored_regions": 2654,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7681581148707517,
+                "region_micro_mean": 0.7685323971146107,
+                "scored_regions": 2654,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7718616823367435,
+                "region_micro_mean": 0.7720426083419587,
+                "scored_regions": 2654,
+                "missing_scores": 0
+              }
+            }
+          },
+          "technique": {
+            "image_macro_mean": 1.8738905775075987,
+            "region_micro_mean": 1.878673700075358,
+            "scored_regions": 2654,
+            "expected_matched_regions": 2654,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 1645,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.7325599811995942,
+                "region_micro_mean": 0.7300501773647778,
+                "scored_regions": 2654,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.8535645349687597,
+                "region_micro_mean": 0.8524965175175541,
+                "scored_regions": 2654,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.8417483321873975,
+                "region_micro_mean": 0.8407375978498236,
+                "scored_regions": 2654,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.8471352425329349,
+                "region_micro_mean": 0.8461078749443233,
+                "scored_regions": 2654,
+                "missing_scores": 0
+              }
+            }
+          },
+          "technical_explanation": {
+            "image_macro_mean": 2.620293819655522,
+            "region_micro_mean": 2.608892238131123,
+            "scored_regions": 2654,
+            "expected_matched_regions": 2654,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 1645,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.34337751722199517,
+                "region_micro_mean": 0.3425555574793207,
+                "scored_regions": 2654,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.7675970028133982,
+                "region_micro_mean": 0.7665316658354201,
+                "scored_regions": 2654,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7578923038104143,
+                "region_micro_mean": 0.7573875704023653,
+                "scored_regions": 2654,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7624980238777526,
+                "region_micro_mean": 0.7617207967255829,
+                "scored_regions": 2654,
+                "missing_scores": 0
+              }
+            }
+          }
+        },
+        "algorithm_metrics_total": {
+          "rouge_l_f1": {
+            "region_micro_mean": 0.48193169517890627,
+            "image_field_macro_mean": 0.48281652496068533,
+            "scored_regions": 7962,
+            "missing_scores": 0
+          },
+          "bertscore_precision": {
+            "region_micro_mean": 0.7983641207173253,
+            "image_field_macro_mean": 0.799079784688737,
+            "scored_regions": 7962,
+            "missing_scores": 0
+          },
+          "bertscore_recall": {
+            "region_micro_mean": 0.7888858551222665,
+            "image_field_macro_mean": 0.7892662502895211,
+            "scored_regions": 7962,
+            "missing_scores": 0
+          },
+          "bertscore_f1": {
+            "region_micro_mean": 0.7932904266706217,
+            "image_field_macro_mean": 0.793831649582477,
+            "scored_regions": 7962,
+            "missing_scores": 0
+          }
+        }
+      },
+      "zifa": {
+        "reference_count": 4540,
+        "hit_count": 1744,
+        "invalid_reference_bbox_count": 0,
+        "recall_micro": 0.3841409691629956,
+        "recall_image_macro": 0.39217099120349896,
+        "matched_iou_mean": 0.7412734126837894,
+        "qualification_rule": "all_required_fields",
+        "qualification_minimum_score": 4,
+        "qualified_analysis_count": 73,
+        "unqualified_hit_count": 1671,
+        "unscored_hit_count": 0,
+        "joint_success_rate": 0.016079295154185023,
+        "qualified_analysis_rate_on_bbox_hits": 0.041857798165137614,
+        "scores": {
+          "evidence": {
+            "image_macro_mean": 2.0405645226049947,
+            "region_micro_mean": 2.00802752293578,
+            "scored_regions": 1744,
+            "expected_matched_regions": 1744,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 1186,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.29623887575757424,
+                "region_micro_mean": 0.2901867053749936,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.734926085052151,
+                "region_micro_mean": 0.7322318013928352,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7301647277493118,
+                "region_micro_mean": 0.7270277670558987,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7322036588200601,
+                "region_micro_mean": 0.7292943135450739,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              }
+            }
+          },
+          "description": {
+            "image_macro_mean": 2.006622902111941,
+            "region_micro_mean": 1.9684633027522935,
+            "scored_regions": 1744,
+            "expected_matched_regions": 1744,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 1186,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.3026969958683862,
+                "region_micro_mean": 0.2931688849888574,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.7129590764197826,
+                "region_micro_mean": 0.7071621390315918,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7129324903988239,
+                "region_micro_mean": 0.7066000620022826,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7121614018409528,
+                "region_micro_mean": 0.7061049249535853,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              }
+            }
+          },
+          "feature": {
+            "image_macro_mean": 1.676724484060066,
+            "region_micro_mean": 1.676605504587156,
+            "scored_regions": 1744,
+            "expected_matched_regions": 1744,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 1186,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.16816180908588,
+                "region_micro_mean": 0.17022153494445436,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.6341161635437423,
+                "region_micro_mean": 0.6336793304002339,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.6268627810177148,
+                "region_micro_mean": 0.6240429911215645,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.6297580548151903,
+                "region_micro_mean": 0.6280210387440176,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              }
+            }
+          },
+          "effect": {
+            "image_macro_mean": 2.190939934152413,
+            "region_micro_mean": 2.1525229357798166,
+            "scored_regions": 1744,
+            "expected_matched_regions": 1744,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 1186,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.3098365731806947,
+                "region_micro_mean": 0.29890201637818076,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.7320491825921377,
+                "region_micro_mean": 0.7254243068602106,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7310704245901315,
+                "region_micro_mean": 0.7249602936208248,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7310902176449541,
+                "region_micro_mean": 0.7247278529706351,
+                "scored_regions": 1744,
+                "missing_scores": 0
+              }
+            }
+          }
+        },
+        "algorithm_metrics_total": {
+          "rouge_l_f1": {
+            "region_micro_mean": 0.2631197854216215,
+            "image_field_macro_mean": 0.26923356347313376,
+            "scored_regions": 6976,
+            "missing_scores": 0
+          },
+          "bertscore_precision": {
+            "region_micro_mean": 0.6996243944212179,
+            "image_field_macro_mean": 0.7035126269019534,
+            "scored_regions": 6976,
+            "missing_scores": 0
+          },
+          "bertscore_recall": {
+            "region_micro_mean": 0.6956577784501426,
+            "image_field_macro_mean": 0.7002576059389954,
+            "scored_regions": 6976,
+            "missing_scores": 0
+          },
+          "bertscore_f1": {
+            "region_micro_mean": 0.697037032553328,
+            "image_field_macro_mean": 0.7013033332802893,
+            "scored_regions": 6976,
+            "missing_scores": 0
+          }
+        }
+      },
+      "all_local_algorithm_metrics": {
+        "rouge_l_f1": {
+          "region_micro_mean": 0.37974720713051835,
+          "scored_regions": 14938,
+          "missing_scores": 0
+        },
+        "bertscore_precision": {
+          "region_micro_mean": 0.752252972595646,
+          "scored_regions": 14938,
+          "missing_scores": 0
+        },
+        "bertscore_recall": {
+          "region_micro_mean": 0.7453486304024421,
+          "scored_regions": 14938,
+          "missing_scores": 0
+        },
+        "bertscore_f1": {
+          "region_micro_mean": 0.7483403880200499,
+          "scored_regions": 14938,
+          "missing_scores": 0
+        }
+      }
+    },
+    "0.5": {
+      "bifa": {
+        "reference_count": 4588,
+        "hit_count": 2258,
+        "invalid_reference_bbox_count": 0,
+        "recall_micro": 0.492153443766347,
+        "recall_image_macro": 0.5245452911724634,
+        "matched_iou_mean": 0.864130889601103,
+        "qualification_rule": "all_required_fields",
+        "qualification_minimum_score": 4,
+        "qualified_analysis_count": 155,
+        "unqualified_hit_count": 2103,
+        "unscored_hit_count": 0,
+        "joint_success_rate": 0.033783783783783786,
+        "qualified_analysis_rate_on_bbox_hits": 0.06864481842338352,
+        "scores": {
+          "visual_evidence": {
+            "image_macro_mean": 3.0623914495657982,
+            "region_micro_mean": 3.079273693534101,
+            "scored_regions": 2258,
+            "expected_matched_regions": 2258,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 1497,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.3863336621696684,
+                "region_micro_mean": 0.3880111414398935,
+                "scored_regions": 2258,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.7810462667426562,
+                "region_micro_mean": 0.7818241487096326,
+                "scored_regions": 2258,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7734972113259411,
+                "region_micro_mean": 0.7744185215504007,
+                "scored_regions": 2258,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7770138133961755,
+                "region_micro_mean": 0.7778659320120478,
+                "scored_regions": 2258,
+                "missing_scores": 0
+              }
+            }
+          },
+          "technique": {
+            "image_macro_mean": 1.9438877755511021,
+            "region_micro_mean": 1.9410983170947742,
+            "scored_regions": 2258,
+            "expected_matched_regions": 2258,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 1497,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.7384101763752221,
+                "region_micro_mean": 0.7366651428035073,
+                "scored_regions": 2258,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.8563289310759092,
+                "region_micro_mean": 0.8553766584744804,
+                "scored_regions": 2258,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.8449640559011408,
+                "region_micro_mean": 0.8442781144692055,
+                "scored_regions": 2258,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.8501584576136337,
+                "region_micro_mean": 0.8493554892795502,
+                "scored_regions": 2258,
+                "missing_scores": 0
+              }
+            }
+          },
+          "technical_explanation": {
+            "image_macro_mean": 2.733656201291472,
+            "region_micro_mean": 2.724534986713906,
+            "scored_regions": 2258,
+            "expected_matched_regions": 2258,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 1497,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.35161893120957355,
+                "region_micro_mean": 0.352030968175649,
+                "scored_regions": 2258,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.7713067007307697,
+                "region_micro_mean": 0.7709574499839803,
+                "scored_regions": 2258,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7618094214144701,
+                "region_micro_mean": 0.7619123883327623,
+                "scored_regions": 2258,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7663158152502115,
+                "region_micro_mean": 0.7661986493293536,
+                "scored_regions": 2258,
+                "missing_scores": 0
+              }
+            }
+          }
+        },
+        "algorithm_metrics_total": {
+          "rouge_l_f1": {
+            "region_micro_mean": 0.4922357508063499,
+            "image_field_macro_mean": 0.492120923251488,
+            "scored_regions": 6774,
+            "missing_scores": 0
+          },
+          "bertscore_precision": {
+            "region_micro_mean": 0.8027194190560311,
+            "image_field_macro_mean": 0.8028939661831117,
+            "scored_regions": 6774,
+            "missing_scores": 0
+          },
+          "bertscore_recall": {
+            "region_micro_mean": 0.7935363414507894,
+            "image_field_macro_mean": 0.7934235628805173,
+            "scored_regions": 6774,
+            "missing_scores": 0
+          },
+          "bertscore_f1": {
+            "region_micro_mean": 0.7978066902069838,
+            "image_field_macro_mean": 0.7978293620866735,
+            "scored_regions": 6774,
+            "missing_scores": 0
+          }
+        }
+      },
+      "zifa": {
+        "reference_count": 4540,
+        "hit_count": 1308,
+        "invalid_reference_bbox_count": 0,
+        "recall_micro": 0.28810572687224667,
+        "recall_image_macro": 0.2794142218290825,
+        "matched_iou_mean": 0.8584477172295785,
+        "qualification_rule": "all_required_fields",
+        "qualification_minimum_score": 4,
+        "qualified_analysis_count": 71,
+        "unqualified_hit_count": 1237,
+        "unscored_hit_count": 0,
+        "joint_success_rate": 0.01563876651982379,
+        "qualified_analysis_rate_on_bbox_hits": 0.054281345565749234,
+        "scores": {
+          "evidence": {
+            "image_macro_mean": 2.1610749185667752,
+            "region_micro_mean": 2.1276758409785934,
+            "scored_regions": 1308,
+            "expected_matched_regions": 1308,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 921,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.30759163346691576,
+                "region_micro_mean": 0.30090035347832783,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.7390089772768397,
+                "region_micro_mean": 0.7362448273266492,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7335943409952616,
+                "region_micro_mean": 0.730904542342603,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7359485218436917,
+                "region_micro_mean": 0.7332284442328532,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              }
+            }
+          },
+          "description": {
+            "image_macro_mean": 2.1852334419109662,
+            "region_micro_mean": 2.1345565749235473,
+            "scored_regions": 1308,
+            "expected_matched_regions": 1308,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 921,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.3151170021741892,
+                "region_micro_mean": 0.30530292934040537,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.7170926653094107,
+                "region_micro_mean": 0.7111795409283507,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7176752599585199,
+                "region_micro_mean": 0.7111792842364821,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7165799285692017,
+                "region_micro_mean": 0.7103813979148136,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              }
+            }
+          },
+          "feature": {
+            "image_macro_mean": 1.7806369887803113,
+            "region_micro_mean": 1.7844036697247707,
+            "scored_regions": 1308,
+            "expected_matched_regions": 1308,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 921,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.19277976467101576,
+                "region_micro_mean": 0.19765773910968032,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.6442463459730838,
+                "region_micro_mean": 0.645606781227873,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.6362332174522782,
+                "region_micro_mean": 0.6352985255898685,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.6394920476134213,
+                "region_micro_mean": 0.6396321585054426,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              }
+            }
+          },
+          "effect": {
+            "image_macro_mean": 2.2443720593557726,
+            "region_micro_mean": 2.21177370030581,
+            "scored_regions": 1308,
+            "expected_matched_regions": 1308,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 921,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.321965535166559,
+                "region_micro_mean": 0.31016338381082353,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.7361230188398917,
+                "region_micro_mean": 0.7288275216606415,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7353959210910014,
+                "region_micro_mean": 0.7289652873161007,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7352789311662689,
+                "region_micro_mean": 0.7284170144468273,
+                "scored_regions": 1308,
+                "missing_scores": 0
+              }
+            }
+          }
+        },
+        "algorithm_metrics_total": {
+          "rouge_l_f1": {
+            "region_micro_mean": 0.27850610143480925,
+            "image_field_macro_mean": 0.2843634838696699,
+            "scored_regions": 5232,
+            "missing_scores": 0
+          },
+          "bertscore_precision": {
+            "region_micro_mean": 0.7054646677858786,
+            "image_field_macro_mean": 0.7091177518498065,
+            "scored_regions": 5232,
+            "missing_scores": 0
+          },
+          "bertscore_recall": {
+            "region_micro_mean": 0.7015869098712636,
+            "image_field_macro_mean": 0.7057246848742652,
+            "scored_regions": 5232,
+            "missing_scores": 0
+          },
+          "bertscore_f1": {
+            "region_micro_mean": 0.7029147537749841,
+            "image_field_macro_mean": 0.7068248572981459,
+            "scored_regions": 5232,
+            "missing_scores": 0
+          }
+        }
+      },
+      "all_local_algorithm_metrics": {
+        "rouge_l_f1": {
+          "region_micro_mean": 0.39909619345903186,
+          "scored_regions": 12006,
+          "missing_scores": 0
+        },
+        "bertscore_precision": {
+          "region_micro_mean": 0.7603375384425514,
+          "scored_regions": 12006,
+          "missing_scores": 0
+        },
+        "bertscore_recall": {
+          "region_micro_mean": 0.7534664242407212,
+          "scored_regions": 12006,
+          "missing_scores": 0
+        },
+        "bertscore_f1": {
+          "region_micro_mean": 0.7564544820267222,
+          "scored_regions": 12006,
+          "missing_scores": 0
+        }
+      }
+    },
+    "0.75": {
+      "bifa": {
+        "reference_count": 4588,
+        "hit_count": 1792,
+        "invalid_reference_bbox_count": 0,
+        "recall_micro": 0.3905841325196164,
+        "recall_image_macro": 0.41366675861146635,
+        "matched_iou_mean": 0.9262921472871408,
+        "qualification_rule": "all_required_fields",
+        "qualification_minimum_score": 4,
+        "qualified_analysis_count": 129,
+        "unqualified_hit_count": 1663,
+        "unscored_hit_count": 0,
+        "joint_success_rate": 0.02811682650392328,
+        "qualified_analysis_rate_on_bbox_hits": 0.07198660714285714,
+        "scores": {
+          "visual_evidence": {
+            "image_macro_mean": 3.155581947743468,
+            "region_micro_mean": 3.1640625,
+            "scored_regions": 1792,
+            "expected_matched_regions": 1792,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 1263,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.3974311548653963,
+                "region_micro_mean": 0.39896029492871465,
+                "scored_regions": 1792,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.7855282986610853,
+                "region_micro_mean": 0.7863037964728262,
+                "scored_regions": 1792,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7783337423435834,
+                "region_micro_mean": 0.7790889839774796,
+                "scored_regions": 1792,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7816691656647838,
+                "region_micro_mean": 0.7824349034843701,
+                "scored_regions": 1792,
+                "missing_scores": 0
+              }
+            }
+          },
+          "technique": {
+            "image_macro_mean": 1.9697149643705463,
+            "region_micro_mean": 1.9765625,
+            "scored_regions": 1792,
+            "expected_matched_regions": 1792,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 1263,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.7420370238052878,
+                "region_micro_mean": 0.7412512023920593,
+                "scored_regions": 1792,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.8583050123908709,
+                "region_micro_mean": 0.8579554842186293,
+                "scored_regions": 1792,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.8470756439160281,
+                "region_micro_mean": 0.846960384598268,
+                "scored_regions": 1792,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.8522080704668135,
+                "region_micro_mean": 0.8520068100520543,
+                "scored_regions": 1792,
+                "missing_scores": 0
+              }
+            }
+          },
+          "technical_explanation": {
+            "image_macro_mean": 2.795526524148852,
+            "region_micro_mean": 2.7896205357142856,
+            "scored_regions": 1792,
+            "expected_matched_regions": 1792,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 1263,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.35820826034425424,
+                "region_micro_mean": 0.35907138061508687,
+                "scored_regions": 1792,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.7739742027522392,
+                "region_micro_mean": 0.7741338476937797,
+                "scored_regions": 1792,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7645089854254009,
+                "region_micro_mean": 0.7648152741936168,
+                "scored_regions": 1792,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7689965291808545,
+                "region_micro_mean": 0.7692332061872419,
+                "scored_regions": 1792,
+                "missing_scores": 0
+              }
+            }
+          }
+        },
+        "algorithm_metrics_total": {
+          "rouge_l_f1": {
+            "region_micro_mean": 0.4997609593119536,
+            "image_field_macro_mean": 0.49922547967164616,
+            "scored_regions": 5376,
+            "missing_scores": 0
+          },
+          "bertscore_precision": {
+            "region_micro_mean": 0.8061310427950784,
+            "image_field_macro_mean": 0.8059358379347318,
+            "scored_regions": 5376,
+            "missing_scores": 0
+          },
+          "bertscore_recall": {
+            "region_micro_mean": 0.7969548809231215,
+            "image_field_macro_mean": 0.7966394572283375,
+            "scored_regions": 5376,
+            "missing_scores": 0
+          },
+          "bertscore_f1": {
+            "region_micro_mean": 0.801224973241222,
+            "image_field_macro_mean": 0.8009579217708173,
+            "scored_regions": 5376,
+            "missing_scores": 0
+          }
+        }
+      },
+      "zifa": {
+        "reference_count": 4540,
+        "hit_count": 947,
+        "invalid_reference_bbox_count": 0,
+        "recall_micro": 0.20859030837004405,
+        "recall_image_macro": 0.1935482087571871,
+        "matched_iou_mean": 0.9533419784565391,
+        "qualification_rule": "all_required_fields",
+        "qualification_minimum_score": 4,
+        "qualified_analysis_count": 56,
+        "unqualified_hit_count": 891,
+        "unscored_hit_count": 0,
+        "joint_success_rate": 0.012334801762114538,
+        "qualified_analysis_rate_on_bbox_hits": 0.059134107708553325,
+        "scores": {
+          "evidence": {
+            "image_macro_mean": 2.2221656976744186,
+            "region_micro_mean": 2.191129883843717,
+            "scored_regions": 947,
+            "expected_matched_regions": 947,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 688,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.31018773878589295,
+                "region_micro_mean": 0.30323513280267383,
+                "scored_regions": 947,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.7404016569343417,
+                "region_micro_mean": 0.7371892366389161,
+                "scored_regions": 947,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7353881237985089,
+                "region_micro_mean": 0.7321180295919038,
+                "scored_regions": 947,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7375341200219222,
+                "region_micro_mean": 0.734297681572319,
+                "scored_regions": 947,
+                "missing_scores": 0
+              }
+            }
+          },
+          "description": {
+            "image_macro_mean": 2.2724806201550387,
+            "region_micro_mean": 2.2302006335797255,
+            "scored_regions": 947,
+            "expected_matched_regions": 947,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 688,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.3222569561488327,
+                "region_micro_mean": 0.31222276249229064,
+                "scored_regions": 947,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.716869497547547,
+                "region_micro_mean": 0.7110878173753603,
+                "scored_regions": 947,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7195136857168504,
+                "region_micro_mean": 0.7134514459838586,
+                "scored_regions": 947,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7173433374209228,
+                "region_micro_mean": 0.7114299443858226,
+                "scored_regions": 947,
+                "missing_scores": 0
+              }
+            }
+          },
+          "feature": {
+            "image_macro_mean": 1.8681928294573644,
+            "region_micro_mean": 1.8521647307286166,
+            "scored_regions": 947,
+            "expected_matched_regions": 947,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 688,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.2271657960066351,
+                "region_micro_mean": 0.22976620104996187,
+                "scored_regions": 947,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.6580212855971483,
+                "region_micro_mean": 0.6589827502919351,
+                "scored_regions": 947,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.6490962161357666,
+                "region_micro_mean": 0.6484705218858426,
+                "scored_regions": 947,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.6527882136195147,
+                "region_micro_mean": 0.6529264535042905,
+                "scored_regions": 947,
+                "missing_scores": 0
+              }
+            }
+          },
+          "effect": {
+            "image_macro_mean": 2.303536821705426,
+            "region_micro_mean": 2.26610348468849,
+            "scored_regions": 947,
+            "expected_matched_regions": 947,
+            "missing_judge_scores": 0,
+            "skipped_missing_reference": 0,
+            "images_fully_scored": 688,
+            "algorithm_metrics": {
+              "rouge_l_f1": {
+                "image_macro_mean": 0.3280038491119573,
+                "region_micro_mean": 0.31697303674441857,
+                "scored_regions": 947,
+                "missing_scores": 0
+              },
+              "bertscore_precision": {
+                "image_macro_mean": 0.7355784818429009,
+                "region_micro_mean": 0.728703529618232,
+                "scored_regions": 947,
+                "missing_scores": 0
+              },
+              "bertscore_recall": {
+                "image_macro_mean": 0.7367019057158352,
+                "region_micro_mean": 0.7307755475059357,
+                "scored_regions": 947,
+                "missing_scores": 0
+              },
+              "bertscore_f1": {
+                "image_macro_mean": 0.7356033870372777,
+                "region_micro_mean": 0.7292097506074996,
+                "scored_regions": 947,
+                "missing_scores": 0
+              }
+            }
+          }
+        },
+        "algorithm_metrics_total": {
+          "rouge_l_f1": {
+            "region_micro_mean": 0.29054928327233626,
+            "image_field_macro_mean": 0.29690358501332953,
+            "scored_regions": 3788,
+            "missing_scores": 0
+          },
+          "bertscore_precision": {
+            "region_micro_mean": 0.7089908334811109,
+            "image_field_macro_mean": 0.7127177304804845,
+            "scored_regions": 3788,
+            "missing_scores": 0
+          },
+          "bertscore_recall": {
+            "region_micro_mean": 0.7062038862418851,
+            "image_field_macro_mean": 0.7101749828417403,
+            "scored_regions": 3788,
+            "missing_scores": 0
+          },
+          "bertscore_f1": {
+            "region_micro_mean": 0.7069659575174829,
+            "image_field_macro_mean": 0.7108172645249093,
+            "scored_regions": 3788,
+            "missing_scores": 0
+          }
+        }
+      },
+      "all_local_algorithm_metrics": {
+        "rouge_l_f1": {
+          "region_micro_mean": 0.4132819295391393,
+          "scored_regions": 9164,
+          "missing_scores": 0
+        },
+        "bertscore_precision": {
+          "region_micro_mean": 0.7659774949031852,
+          "scored_regions": 9164,
+          "missing_scores": 0
+        },
+        "bertscore_recall": {
+          "region_micro_mean": 0.759442357150476,
+          "scored_regions": 9164,
+          "missing_scores": 0
+        },
+        "bertscore_f1": {
+          "region_micro_mean": 0.7622623857727013,
+          "scored_regions": 9164,
+          "missing_scores": 0
+        }
+      }
+    }
+  },
+  "overall": {
+    "overall_comment_mean": 2.6307129798903106,
+    "scored_images": 2188,
+    "skipped_missing_reference": 0,
+    "unscored_images": 0,
+    "algorithm_metrics": {
+      "rouge_l_f1": {
+        "mean": 0.2559770338133152,
+        "scored_images": 2188,
+        "missing_scores": 0
+      },
+      "bertscore_precision": {
+        "mean": 0.7495749348968649,
+        "scored_images": 2188,
+        "missing_scores": 0
+      },
+      "bertscore_recall": {
+        "mean": 0.7401539133333854,
+        "scored_images": 2188,
+        "missing_scores": 0
+      },
+      "bertscore_f1": {
+        "mean": 0.7447555589828456,
+        "scored_images": 2188,
+        "missing_scores": 0
+      }
+    }
+  }
+}

sft/qwen3vl_sft_v16/inference_best/inference_config.json

@@ -0,0 +1,13 @@
+{
+  "model": "/home/jca/models/Qwen3-VL-8B-Instruct",
+  "adapters": [
+    "/home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611/checkpoint-5000"
+  ],
+  "dataset_sha256": "958282611539a7a037955b038d02b55b96c161923ad062b11c500c5988456e2e",
+  "max_new_tokens": 3072,
+  "temperature": 0.0,
+  "top_p": 1.0,
+  "dtype": "bfloat16",
+  "image_max_token_num": 512,
+  "image_min_token_num": 4
+}

sft/qwen3vl_sft_v16/inference_best/inference_summary.json

@@ -0,0 +1,29 @@
+{
+  "model": "/home/jca/models/Qwen3-VL-8B-Instruct",
+  "adapters": [
+    "/home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611/checkpoint-5000"
+  ],
+  "dataset": "/home/jca/calligraphy_appreciation/data/sft/test_inference.jsonl",
+  "samples": 2188,
+  "processed": 2188,
+  "resumed_success": 0,
+  "new_processed": 2188,
+  "success": 2188,
+  "statuses": {
+    "ok": 2188
+  },
+  "remaining_or_failed": 0,
+  "interrupted": false,
+  "fatal_error": null,
+  "elapsed_seconds": 6523.841,
+  "gpus": [
+    "0",
+    "1"
+  ],
+  "workers": 2,
+  "batch_size_per_gpu": 10,
+  "temperature": 0.0,
+  "top_p": 1.0,
+  "max_new_tokens": 3072,
+  "latency_note": "该样本所在批次的耗时，不应在批内累加"
+}

sft/qwen3vl_sft_v16/inference_best/worker_logs/worker_0.log

@@ -0,0 +1,137 @@
+GPU=0, batch_size=10, model=/home/jca/models/Qwen3-VL-8B-Instruct
+[WARNING:swift] Loading model from '/home/jca/models/Qwen3-VL-8B-Instruct' with trust_remote_code=True. This will execute custom Python code shipped in the model repository (e.g. configuration_*.py, modeling_*.py, tokenization_*.py). Only proceed if you trust the source of this model.
+[transformers] `torch_dtype` is deprecated! Use `dtype` instead!
+[WARNING:swift] Loading config from '/home/jca/models/Qwen3-VL-8B-Instruct' with trust_remote_code=True. Custom Python code from the model repository will be executed. Only use models from sources you trust.
+[WARNING:swift] Loading tokenizer/processor from '/home/jca/models/Qwen3-VL-8B-Instruct' with trust_remote_code=True. Custom Python code from the model repository will be executed. Only use models from sources you trust.
+[INFO:swift] Setting max_ratio: 200. You can adjust this hyperparameter through the environment variable: `MAX_RATIO`.
+[INFO:swift] Setting frame_factor: 2. You can adjust this hyperparameter through the environment variable: `FRAME_FACTOR`.
+[INFO:swift] Setting fps: 2.0. You can adjust this hyperparameter through the environment variable: `FPS`.
+[INFO:swift] Setting fps_min_frames: 4. You can adjust this hyperparameter through the environment variable: `FPS_MIN_FRAMES`.
+[INFO:swift] Setting fps_max_frames: 768. You can adjust this hyperparameter through the environment variable: `FPS_MAX_FRAMES`.
+[INFO:swift] Using environment variable `IMAGE_MAX_TOKEN_NUM`, Setting image_max_token_num: 512.
+[INFO:swift] Using environment variable `IMAGE_MIN_TOKEN_NUM`, Setting image_min_token_num: 4.
+[INFO:swift] Setting spatial_merge_size: 2. You can adjust this hyperparameter through the environment variable: `SPATIAL_MERGE_SIZE`.
+[INFO:swift] Setting video_max_token_num: 768. You can adjust this hyperparameter through the environment variable: `VIDEO_MAX_TOKEN_NUM`.
+[INFO:swift] Setting video_min_token_num: 128. You can adjust this hyperparameter through the environment variable: `VIDEO_MIN_TOKEN_NUM`.
+[INFO:swift] model_kwargs: {'device_map': 'cuda:0', 'dtype': torch.bfloat16}
+
+[INFO:swift] Create the template for the infer_engine
+[INFO:swift] Successfully registered `/opt/conda_envs/jca/qwen3vl/lib/python3.12/site-packages/swift/dataset/data/dataset_info.json`.
+[INFO:swift] Successfully loaded /home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611/checkpoint-5000/args.json.
+[INFO:swift] default_system: None
+[INFO:swift] max_length: 262144
+[INFO:swift] response_prefix: None
+[INFO:swift] agent_template: hermes
+[INFO:swift] norm_bbox: norm1000
+[INFO:swift] Setting ROOT_IMAGE_DIR: None. You can adjust this hyperparameter through the environment variable: `ROOT_IMAGE_DIR`.
+[INFO:swift] Setting QWENVL_BBOX_FORMAT: legacy. You can adjust this hyperparameter through the environment variable: `QWENVL_BBOX_FORMAT`.
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/8 [00:00<?, ?it/s]

sft/qwen3vl_sft_v16/inference_best/worker_logs/worker_1.log

@@ -0,0 +1,136 @@
+GPU=1, batch_size=10, model=/home/jca/models/Qwen3-VL-8B-Instruct
+[WARNING:swift] Loading model from '/home/jca/models/Qwen3-VL-8B-Instruct' with trust_remote_code=True. This will execute custom Python code shipped in the model repository (e.g. configuration_*.py, modeling_*.py, tokenization_*.py). Only proceed if you trust the source of this model.
+[transformers] `torch_dtype` is deprecated! Use `dtype` instead!
+[WARNING:swift] Loading config from '/home/jca/models/Qwen3-VL-8B-Instruct' with trust_remote_code=True. Custom Python code from the model repository will be executed. Only use models from sources you trust.
+[WARNING:swift] Loading tokenizer/processor from '/home/jca/models/Qwen3-VL-8B-Instruct' with trust_remote_code=True. Custom Python code from the model repository will be executed. Only use models from sources you trust.
+[INFO:swift] Setting max_ratio: 200. You can adjust this hyperparameter through the environment variable: `MAX_RATIO`.
+[INFO:swift] Setting frame_factor: 2. You can adjust this hyperparameter through the environment variable: `FRAME_FACTOR`.
+[INFO:swift] Setting fps: 2.0. You can adjust this hyperparameter through the environment variable: `FPS`.
+[INFO:swift] Setting fps_min_frames: 4. You can adjust this hyperparameter through the environment variable: `FPS_MIN_FRAMES`.
+[INFO:swift] Setting fps_max_frames: 768. You can adjust this hyperparameter through the environment variable: `FPS_MAX_FRAMES`.
+[INFO:swift] Using environment variable `IMAGE_MAX_TOKEN_NUM`, Setting image_max_token_num: 512.
+[INFO:swift] Using environment variable `IMAGE_MIN_TOKEN_NUM`, Setting image_min_token_num: 4.
+[INFO:swift] Setting spatial_merge_size: 2. You can adjust this hyperparameter through the environment variable: `SPATIAL_MERGE_SIZE`.
+[INFO:swift] Setting video_max_token_num: 768. You can adjust this hyperparameter through the environment variable: `VIDEO_MAX_TOKEN_NUM`.
+[INFO:swift] Setting video_min_token_num: 128. You can adjust this hyperparameter through the environment variable: `VIDEO_MIN_TOKEN_NUM`.
+[INFO:swift] model_kwargs: {'device_map': 'cuda:0', 'dtype': torch.bfloat16}
+
+[INFO:swift] Create the template for the infer_engine
+[INFO:swift] Successfully registered `/opt/conda_envs/jca/qwen3vl/lib/python3.12/site-packages/swift/dataset/data/dataset_info.json`.
+[INFO:swift] Successfully loaded /home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611/checkpoint-5000/args.json.
+[INFO:swift] default_system: None
+[INFO:swift] max_length: 262144
+[INFO:swift] response_prefix: None
+[INFO:swift] agent_template: hermes
+[INFO:swift] norm_bbox: norm1000
+[INFO:swift] Setting ROOT_IMAGE_DIR: None. You can adjust this hyperparameter through the environment variable: `ROOT_IMAGE_DIR`.
+[INFO:swift] Setting QWENVL_BBOX_FORMAT: legacy. You can adjust this hyperparameter through the environment variable: `QWENVL_BBOX_FORMAT`.
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]
+
  0%|          | 0/10 [00:00<?, ?it/s]

sft/qwen3vl_sft_v16/preflight/preflight_summary.json

@@ -0,0 +1,135 @@
+{
+  "model": "/home/jca/models/Qwen3-VL-8B-Instruct",
+  "backend": "swift",
+  "max_length": 4096,
+  "errors": 0,
+  "error_examples": [],
+  "datasets": {
+    "train": {
+      "path": "/home/jca/calligraphy_appreciation/data/sft/train_multitask.jsonl",
+      "rows": 26992,
+      "invalid_rows": 0,
+      "encoded_rows": 26992,
+      "length_statistics": {
+        "samples": 26992,
+        "min": 175,
+        "max": 3104,
+        "mean": 1095.4009706579727,
+        "p50": 1312.0,
+        "p90": 1676.0,
+        "p95": 1841.449999999997,
+        "p99": 2252.1800000000003,
+        "configured_max_length": 4096,
+        "over_configured_max_length": 0,
+        "candidate_limits": {
+          "2048": {
+            "over": 643,
+            "kept_ratio": 0.9761781268524007
+          },
+          "4096": {
+            "over": 0,
+            "kept_ratio": 1.0
+          },
+          "8192": {
+            "over": 0,
+            "kept_ratio": 1.0
+          },
+          "12288": {
+            "over": 0,
+            "kept_ratio": 1.0
+          }
+        },
+        "prompt_token_statistics": {
+          "count": 26992,
+          "min": 145,
+          "max": 836,
+          "mean": 625.7000592768228,
+          "p50": 756.0,
+          "p95": 836.0,
+          "p99": 836.0
+        },
+        "assistant_token_statistics": {
+          "count": 26992,
+          "min": 14,
+          "max": 2428,
+          "mean": 469.70091138115,
+          "p50": 530.0,
+          "p95": 1062.449999999997,
+          "p99": 1469.2700000000004
+        },
+        "image_token_statistics": {
+          "count": 26992,
+          "min": 32,
+          "max": 256,
+          "mean": 204.4463544754001,
+          "p50": 208.0,
+          "p95": 256.0,
+          "p99": 256.0
+        }
+      }
+    },
+    "validation": {
+      "path": "/home/jca/calligraphy_appreciation/data/sft/validation.jsonl",
+      "rows": 2199,
+      "invalid_rows": 0,
+      "encoded_rows": 2199,
+      "length_statistics": {
+        "samples": 2199,
+        "min": 782,
+        "max": 2712,
+        "mean": 1485.286948613006,
+        "p50": 1460.0,
+        "p90": 1778.2,
+        "p95": 1954.2999999999997,
+        "p99": 2404.08,
+        "configured_max_length": 4096,
+        "over_configured_max_length": 0,
+        "candidate_limits": {
+          "2048": {
+            "over": 76,
+            "kept_ratio": 0.9654388358344702
+          },
+          "4096": {
+            "over": 0,
+            "kept_ratio": 1.0
+          },
+          "8192": {
+            "over": 0,
+            "kept_ratio": 1.0
+          },
+          "12288": {
+            "over": 0,
+            "kept_ratio": 1.0
+          }
+        },
+        "prompt_token_statistics": {
+          "count": 2199,
+          "min": 612,
+          "max": 836,
+          "mean": 784.7403365165984,
+          "p50": 788.0,
+          "p95": 836.0,
+          "p99": 836.0
+        },
+        "assistant_token_statistics": {
+          "count": 2199,
+          "min": 108,
+          "max": 1908,
+          "mean": 700.5466120964074,
+          "p50": 671.0,
+          "p95": 1177.1,
+          "p99": 1612.02
+        },
+        "image_token_statistics": {
+          "count": 2199,
+          "min": 32,
+          "max": 256,
+          "mean": 204.74033651659846,
+          "p50": 208.0,
+          "p95": 256.0,
+          "p99": 256.0
+        }
+      }
+    }
+  }
+}

sft/qwen3vl_sft_v16/v0-20261028-121611/args.json

@@ -0,0 +1,359 @@
+{
+  "output_dir": "/home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611",
+  "per_device_train_batch_size": 2,
+  "num_train_epochs": 4.0,
+  "max_steps": -1,
+  "learning_rate": 0.0001,
+  "lr_scheduler_type": "cosine",
+  "lr_scheduler_kwargs": null,
+  "warmup_steps": 0,
+  "optim": "adamw_torch_fused",
+  "optim_args": null,
+  "weight_decay": 0.1,
+  "adam_beta1": 0.9,
+  "adam_beta2": 0.95,
+  "adam_epsilon": 1e-08,
+  "optim_target_modules": null,
+  "gradient_accumulation_steps": 4,
+  "average_tokens_across_devices": true,
+  "max_grad_norm": 1.0,
+  "label_smoothing_factor": 0.0,
+  "bf16": true,
+  "fp16": false,
+  "bf16_full_eval": false,
+  "fp16_full_eval": false,
+  "tf32": null,
+  "gradient_checkpointing": true,
+  "gradient_checkpointing_kwargs": null,
+  "torch_compile": false,
+  "torch_compile_backend": null,
+  "torch_compile_mode": null,
+  "use_liger_kernel": false,
+  "liger_kernel_config": null,
+  "use_cache": false,
+  "neftune_noise_alpha": null,
+  "torch_empty_cache_steps": null,
+  "auto_find_batch_size": false,
+  "logging_strategy": "steps",
+  "logging_steps": 5,
+  "logging_first_step": true,
+  "log_on_each_node": true,
+  "logging_nan_inf_filter": true,
+  "include_num_input_tokens_seen": false,
+  "log_level": "passive",
+  "log_level_replica": "warning",
+  "disable_tqdm": null,
+  "report_to": [
+    "tensorboard"
+  ],
+  "run_name": "/home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611",
+  "project": "huggingface",
+  "trackio_space_id": null,
+  "trackio_bucket_id": null,
+  "trackio_static_space_id": null,
+  "eval_strategy": "steps",
+  "eval_steps": 200.0,
+  "eval_delay": 0,
+  "per_device_eval_batch_size": 2,
+  "prediction_loss_only": false,
+  "eval_on_start": false,
+  "eval_do_concat_batches": true,
+  "eval_use_gather_object": false,
+  "eval_accumulation_steps": null,
+  "include_for_metrics": [],
+  "batch_eval_metrics": false,
+  "save_only_model": false,
+  "save_strategy": "steps",
+  "save_steps": 200.0,
+  "save_on_each_node": false,
+  "save_total_limit": 3,
+  "enable_jit_checkpoint": false,
+  "push_to_hub": false,
+  "hub_token": null,
+  "hub_private_repo": null,
+  "hub_model_id": null,
+  "hub_strategy": "every_save",
+  "hub_always_push": false,
+  "hub_revision": null,
+  "load_best_model_at_end": true,
+  "metric_for_best_model": "loss",
+  "greater_is_better": false,
+  "ignore_data_skip": false,
+  "restore_callback_states_from_checkpoint": false,
+  "full_determinism": false,
+  "seed": 42,
+  "data_seed": 42,
+  "use_cpu": false,
+  "accelerator_config": {
+    "dispatch_batches": false
+  },
+  "parallelism_config": null,
+  "dataloader_drop_last": false,
+  "dataloader_num_workers": 2,
+  "dataloader_pin_memory": true,
+  "dataloader_persistent_workers": true,
+  "dataloader_prefetch_factor": null,
+  "dataloader_multiprocessing_context": null,
+  "dataloader_in_order": true,
+  "remove_unused_columns": true,
+  "label_names": null,
+  "train_sampling_strategy": "random",
+  "length_column_name": "length",
+  "ddp_find_unused_parameters": null,
+  "ddp_bucket_cap_mb": null,
+  "ddp_broadcast_buffers": null,
+  "ddp_static_graph": null,
+  "ddp_backend": null,
+  "ddp_timeout": 18000000,
+  "fsdp": [],
+  "fsdp_config": null,
+  "deepspeed": null,
+  "debug": null,
+  "skip_memory_metrics": true,
+  "do_train": false,
+  "do_eval": false,
+  "do_predict": false,
+  "resume_from_checkpoint": null,
+  "local_rank": 0,
+  "sortish_sampler": false,
+  "predict_with_generate": false,
+  "generation_max_length": null,
+  "generation_num_beams": null,
+  "generation_config": null,
+  "tuner_backend": "peft",
+  "vit_gradient_checkpointing": true,
+  "router_aux_loss_coef": 0.0,
+  "enable_dft_loss": false,
+  "enable_channel_loss": false,
+  "safe_serialization": true,
+  "max_shard_size": "5GB",
+  "check_model": true,
+  "acc_strategy": "token",
+  "train_dataloader_shuffle": true,
+  "group_by_length": false,
+  "max_epochs": null,
+  "aligner_lr": null,
+  "vit_lr": 1e-05,
+  "use_logits_to_keep": null,
+  "ds3_gather_for_generation": true,
+  "resume_only_model": false,
+  "optimizer": null,
+  "loss_type": null,
+  "mrl_dims": null,
+  "eval_metric": null,
+  "callbacks": [],
+  "early_stop_interval": null,
+  "eval_use_evalscope": false,
+  "eval_dataset": [],
+  "eval_dataset_args": null,
+  "eval_limit": null,
+  "eval_generation_config": null,
+  "extra_eval_args": null,
+  "tuner_type": "lora",
+  "use_galore": false,
+  "galore_target_modules": null,
+  "galore_rank": 128,
+  "galore_update_proj_gap": 50,
+  "galore_scale": 1.0,
+  "galore_proj_type": "std",
+  "galore_optim_per_parameter": false,
+  "galore_with_embedding": false,
+  "galore_quantization": false,
+  "galore_proj_quant": false,
+  "galore_proj_bits": 4,
+  "galore_proj_group_size": 256,
+  "galore_cos_threshold": 0.4,
+  "galore_gamma_proj": 2,
+  "galore_queue_size": 5,
+  "lisa_activated_layers": 0,
+  "lisa_step_interval": 20,
+  "use_flash_ckpt": false,
+  "logging_dir": "/home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611/runs",
+  "warmup_ratio": 0.05,
+  "use_ray": false,
+  "ray_exp_name": null,
+  "device_groups": null,
+  "model": "/home/jca/models/Qwen3-VL-8B-Instruct",
+  "model_type": "qwen3_vl",
+  "model_revision": null,
+  "task_type": "causal_lm",
+  "torch_dtype": "bfloat16",
+  "attn_impl": null,
+  "experts_impl": null,
+  "new_special_tokens": [],
+  "num_labels": null,
+  "problem_type": null,
+  "rope_scaling": null,
+  "device_map": null,
+  "max_memory": {},
+  "max_model_len": null,
+  "local_repo_path": null,
+  "init_strategy": null,
+  "template": "qwen3_vl",
+  "system": null,
+  "max_length": 4096,
+  "truncation_strategy": "delete",
+  "max_pixels": null,
+  "agent_template": null,
+  "norm_bbox": null,
+  "use_chat_template": true,
+  "padding_side": "right",
+  "padding_free": false,
+  "loss_scale": "default",
+  "sequence_parallel_size": 1,
+  "is_binary_loss_scale": null,
+  "template_backend": "swift",
+  "response_prefix": null,
+  "enable_thinking": null,
+  "preserve_thinking": null,
+  "add_non_thinking_prefix": true,
+  "disable_ignore_empty_think": false,
+  "dataset": [
+    "/home/jca/calligraphy_appreciation/data/sft/train_multitask.jsonl"
+  ],
+  "val_dataset": [
+    "/home/jca/calligraphy_appreciation/data/sft/validation.jsonl"
+  ],
+  "cached_dataset": [],
+  "cached_val_dataset": [],
+  "split_dataset_ratio": 0.0,
+  "dataset_num_proc": 2,
+  "load_from_cache_file": false,
+  "dataset_shuffle": true,
+  "val_dataset_shuffle": false,
+  "streaming": false,
+  "interleave_prob": null,
+  "stopping_strategy": "first_exhausted",
+  "shuffle_buffer_size": 1000,
+  "download_mode": "reuse_dataset_if_exists",
+  "columns": {},
+  "strict": true,
+  "disable_auto_column_mapping": false,
+  "model_name": null,
+  "model_author": null,
+  "custom_dataset_info": [],
+  "decision_drop_instructions": 0.0,
+  "decision_subsample_options": 0,
+  "decision_state_as_text": 0.0,
+  "decision_shuffle_fields": false,
+  "quant_method": "bnb",
+  "quant_bits": 4,
+  "hqq_axis": null,
+  "bnb_4bit_compute_dtype": "bfloat16",
+  "bnb_4bit_quant_type": "nf4",
+  "bnb_4bit_use_double_quant": true,
+  "bnb_4bit_quant_storage": null,
+  "max_new_tokens": 64,
+  "temperature": 0.0,
+  "top_k": null,
+  "top_p": null,
+  "repetition_penalty": null,
+  "num_beams": 1,
+  "stream": false,
+  "stop_words": [],
+  "logprobs": false,
+  "top_logprobs": null,
+  "structured_outputs_regex": null,
+  "adapters": [],
+  "external_plugins": [],
+  "custom_register_path": [],
+  "model_kwargs": {},
+  "enable_npu_model_patch": true,
+  "load_args": false,
+  "load_data_args": false,
+  "packing": false,
+  "packing_length": null,
+  "packing_num_proc": 1,
+  "packing_strategy": "binpack",
+  "lazy_tokenize": true,
+  "use_hf": false,
+  "ignore_args_error": false,
+  "use_swift_lora": false,
+  "freeze_parameters": [],
+  "freeze_parameters_regex": null,
+  "freeze_parameters_ratio": 0.0,
+  "trainable_parameters": [],
+  "trainable_parameters_regex": null,
+  "freeze_llm": false,
+  "freeze_vit": false,
+  "freeze_aligner": true,
+  "target_modules": [
+    "all-linear"
+  ],
+  "target_regex": null,
+  "target_parameters": null,
+  "modules_to_save": [],
+  "lora_rank": 128,
+  "lora_alpha": 256.0,
+  "lora_dropout": 0.05,
+  "lora_bias": "none",
+  "lora_dtype": null,
+  "lorap_lr_ratio": null,
+  "use_rslora": false,
+  "use_dora": false,
+  "lora_ga_batch_size": 2,
+  "lora_ga_iters": 2,
+  "lora_ga_max_length": 1024,
+  "lora_ga_direction": "ArB2r",
+  "lora_ga_scale": "stable",
+  "lora_ga_stable_gamma": 16,
+  "init_weights": true,
+  "fourier_n_frequency": 2000,
+  "fourier_scaling": 300.0,
+  "boft_block_size": 4,
+  "boft_block_num": 0,
+  "boft_n_butterfly_factor": 1,
+  "boft_dropout": 0.0,
+  "vera_rank": 256,
+  "vera_projection_prng_key": 0,
+  "vera_dropout": 0.0,
+  "vera_d_initial": 0.1,
+  "adapter_act": "gelu",
+  "adapter_length": 128,
+  "adalora_target_r": 8,
+  "adalora_init_r": 12,
+  "adalora_tinit": 0,
+  "adalora_tfinal": 0,
+  "adalora_deltaT": 1,
+  "adalora_beta1": 0.85,
+  "adalora_beta2": 0.85,
+  "adalora_orth_reg_weight": 0.5,
+  "llamapro_num_new_blocks": 4,
+  "llamapro_num_groups": null,
+  "reft_layer_key": null,
+  "reft_layers": null,
+  "reft_rank": 4,
+  "reft_intervention_type": "LoreftIntervention",
+  "reft_args": null,
+  "swanlab_token": null,
+  "swanlab_project": "ms-swift",
+  "swanlab_workspace": null,
+  "swanlab_exp_name": null,
+  "swanlab_notification_method": null,
+  "swanlab_webhook_url": null,
+  "swanlab_secret": null,
+  "swanlab_sender_email": null,
+  "swanlab_receiver_email": null,
+  "swanlab_smtp_server": null,
+  "swanlab_smtp_port": null,
+  "swanlab_email_language": "zh",
+  "swanlab_mode": "cloud",
+  "add_version": true,
+  "create_checkpoint_symlink": true,
+  "zero_hpz_partition_size": null,
+  "deepspeed_autotp_size": null,
+  "swift_version": "4.6.0",
+  "ckpt_dir": null,
+  "rank": 0,
+  "global_world_size": 2,
+  "local_world_size": 2,
+  "model_suffix": "Qwen3-VL-8B-Instruct",
+  "model_info": "ModelInfo(model_type='qwen3_vl', model_dir='/home/jca/models/Qwen3-VL-8B-Instruct', torch_dtype=torch.bfloat16, max_model_len=262144, quant_method='bnb', quant_bits=4, rope_scaling=None, is_moe_model=False, is_multimodal=True, config=None, task_type='causal_lm', num_labels=None)",
+  "model_meta": "ModelMeta(model_type='qwen3_vl', model_groups=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen3-VL-2B-Instruct', hf_model_id='Qwen/Qwen3-VL-2B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-2B-Thinking', hf_model_id='Qwen/Qwen3-VL-2B-Thinking', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-2B-Instruct-FP8', hf_model_id='Qwen/Qwen3-VL-2B-Instruct-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-2B-Thinking-FP8', hf_model_id='Qwen/Qwen3-VL-2B-Thinking-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-4B-Instruct', hf_model_id='Qwen/Qwen3-VL-4B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-4B-Thinking', hf_model_id='Qwen/Qwen3-VL-4B-Thinking', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-4B-Instruct-FP8', hf_model_id='Qwen/Qwen3-VL-4B-Instruct-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-4B-Thinking-FP8', hf_model_id='Qwen/Qwen3-VL-4B-Thinking-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-8B-Instruct', hf_model_id='Qwen/Qwen3-VL-8B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-8B-Thinking', hf_model_id='Qwen/Qwen3-VL-8B-Thinking', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-8B-Instruct-FP8', hf_model_id='Qwen/Qwen3-VL-8B-Instruct-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-8B-Thinking-FP8', hf_model_id='Qwen/Qwen3-VL-8B-Thinking-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-32B-Instruct', hf_model_id='Qwen/Qwen3-VL-32B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-32B-Thinking', hf_model_id='Qwen/Qwen3-VL-32B-Thinking', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-32B-Instruct-FP8', hf_model_id='Qwen/Qwen3-VL-32B-Instruct-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-32B-Thinking-FP8', hf_model_id='Qwen/Qwen3-VL-32B-Thinking-FP8', model_path=None, ms_revision=None, hf_revision=None)], template='qwen3_vl', ignore_patterns=None, requires=None, tags=[])], loader=<class 'swift.model.models.qwen.Qwen3VLLoader'>, template='qwen3_vl', model_arch=MultiModelKeys(arch_name='qwen3_vl', embedding=None, module_list=None, lm_head=None, q_proj=None, k_proj=None, v_proj=None, o_proj=None, attention=None, mlp='model.language_model.layers.{}.mlp', down_proj=None, qkv_proj=None, qk_proj=None, qa_proj=None, qb_proj=None, kv_proj=None, kva_proj=None, kvb_proj=None, language_model=['model.language_model', 'lm_head'], aligner=['model.visual.merger', 'model.visual.deepstack_merger_list'], vision_tower=['model.visual'], generator=[]), mcore_model_type=None, architectures=['Qwen3VLForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=['transformers>=4.57', 'qwen_vl_utils>=0.0.14', 'decord'], tags=[])",
+  "model_dir": "/home/jca/models/Qwen3-VL-8B-Instruct",
+  "template_meta": "QwenTemplateMeta(template_type='qwen3_vl', prefix=[], prompt=['<|im_start|>user\\n{{QUERY}}<|im_end|>\\n<|im_start|>assistant\\n'], chat_sep=['<|im_end|>\\n'], suffix=['<|im_end|>\\n'], template_cls=<class 'swift.template.templates.qwen.Qwen3VLTemplate'>, system_prefix=['<|im_start|>system\\n{{SYSTEM}}<|im_end|>\\n'], default_system=None, auto_add_bos=False, stop_words=['<|endoftext|>'], agent_template='hermes', is_thinking=False, thinking_prefix='<think>\\n', non_thinking_prefix='', preserve_thinking=None, history_thinking_prefix='')",
+  "_val_dataset_exists": true,
+  "hub": "<class 'swift.hub.hub.MSHub'>",
+  "evaluation_strategy": "steps",
+  "training_args": "Seq2SeqTrainingArguments(output_dir='/home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611', per_device_train_batch_size=2, num_train_epochs=4.0, max_steps=-1, learning_rate=0.0001, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_steps=0.05, optim=<OptimizerNames.ADAMW_TORCH_FUSED: 'adamw_torch_fused'>, optim_args=None, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, optim_target_modules=None, gradient_accumulation_steps=4, average_tokens_across_devices=None, max_grad_norm=1.0, label_smoothing_factor=0.0, bf16=True, fp16=False, bf16_full_eval=False, fp16_full_eval=False, tf32=None, gradient_checkpointing=True, gradient_checkpointing_kwargs=None, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, use_liger_kernel=False, liger_kernel_config=None, use_cache=False, neftune_noise_alpha=None, torch_empty_cache_steps=None, auto_find_batch_size=False, logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_steps=5, logging_first_step=True, log_on_each_node=True, logging_nan_inf_filter=True, include_num_input_tokens_seen=None, log_level='passive', log_level_replica='warning', disable_tqdm=False, report_to=['tensorboard'], run_name='/home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611', project='huggingface', trackio_space_id=None, trackio_bucket_id=None, trackio_static_space_id=None, eval_strategy=<IntervalStrategy.STEPS: 'steps'>, eval_steps=200, eval_delay=0, per_device_eval_batch_size=2, prediction_loss_only=False, eval_on_start=False, eval_do_concat_batches=True, eval_use_gather_object=False, eval_accumulation_steps=None, include_for_metrics=[], batch_eval_metrics=False, save_only_model=False, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=200, save_on_each_node=False, save_total_limit=3, enable_jit_checkpoint=False, push_to_hub=False, hub_token=None, hub_private_repo=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_always_push=False, hub_revision=None, load_best_model_at_end=True, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, restore_callback_states_from_checkpoint=False, full_determinism=False, seed=42, data_seed=42, use_cpu=False, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), parallelism_config=None, dataloader_drop_last=False, dataloader_num_workers=2, dataloader_pin_memory=True, dataloader_persistent_workers=True, dataloader_prefetch_factor=2, dataloader_multiprocessing_context=None, dataloader_in_order=True, remove_unused_columns=False, label_names=None, train_sampling_strategy='random', length_column_name='length', ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, ddp_static_graph=None, ddp_backend=None, ddp_timeout=18000000, fsdp=[], fsdp_config=None, deepspeed=None, debug=[], skip_memory_metrics=True, do_train=False, do_eval=True, do_predict=False, resume_from_checkpoint=None, local_rank=0, sortish_sampler=False, predict_with_generate=False, generation_max_length=None, generation_num_beams=None, generation_config=None, tuner_backend='peft', vit_gradient_checkpointing=True, router_aux_loss_coef=0.0, enable_dft_loss=False, enable_channel_loss=False, safe_serialization=True, max_shard_size='5GB', check_model=True, acc_strategy='token', train_dataloader_shuffle=True, group_by_length=False, max_epochs=None, aligner_lr=None, vit_lr=1e-05, use_logits_to_keep=None, ds3_gather_for_generation=True, resume_only_model=False, optimizer='multimodal', loss_type=None, mrl_dims=None, eval_metric=None, callbacks=[], early_stop_interval=None, eval_use_evalscope=False, eval_dataset=[], eval_dataset_args=None, eval_limit=None, eval_generation_config=None, extra_eval_args=None, tuner_type='lora', use_galore=False, galore_target_modules=None, galore_rank=128, galore_update_proj_gap=50, galore_scale=1.0, galore_proj_type='std', galore_optim_per_parameter=False, galore_with_embedding=False, galore_quantization=False, galore_proj_quant=False, galore_proj_bits=4, galore_proj_group_size=256, galore_cos_threshold=0.4, galore_gamma_proj=2, galore_queue_size=5, lisa_activated_layers=0, lisa_step_interval=20, use_flash_ckpt=False, logging_dir='/home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611/runs', warmup_ratio=0.0)"
+}

sft/qwen3vl_sft_v16/v0-20261028-121611/checkpoint-5000/README.md

@@ -0,0 +1,207 @@
+---
+base_model: /home/jca/models/Qwen3-VL-8B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:/home/jca/models/Qwen3-VL-8B-Instruct
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.20.0

sft/qwen3vl_sft_v16/v0-20261028-121611/checkpoint-5000/adapter_config.json

@@ -0,0 +1,42 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/home/jca/models/Qwen3-VL-8B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 256.0,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": [],
+  "monteclora_config": null,
+  "peft_type": "LORA",
+  "peft_version": "0.20.0",
+  "qalora_group_size": 16,
+  "r": 128,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": "^(model\\.language_model(?=\\.).*\\.(o_proj|gate_proj|down_proj|v_proj|up_proj|k_proj|q_proj)|(?!(model.visual.merger|model.visual.deepstack_merger_list))model\\.visual(?=\\.).*\\.(linear_fc1|qkv|linear_fc2|attn.proj))$",
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false,
+  "velora_config": null
+}

sft/qwen3vl_sft_v16/v0-20261028-121611/checkpoint-5000/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02d5a3f359aed69d20b54d6ef96e6ec0dcabf6b7035007e1c5a1ea81feb25cb6
+size 1643209400

sft/qwen3vl_sft_v16/v0-20261028-121611/checkpoint-5000/additional_config.json

@@ -0,0 +1 @@
+{"lora_dtype": null, "lorap_lr_ratio": null, "lorap_emb_lr": 1e-06}

sft/qwen3vl_sft_v16/v0-20261028-121611/checkpoint-5000/args.json

@@ -0,0 +1,359 @@
+{
+  "output_dir": "/home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611",
+  "per_device_train_batch_size": 2,
+  "num_train_epochs": 4.0,
+  "max_steps": -1,
+  "learning_rate": 0.0001,
+  "lr_scheduler_type": "cosine",
+  "lr_scheduler_kwargs": null,
+  "warmup_steps": 0,
+  "optim": "adamw_torch_fused",
+  "optim_args": null,
+  "weight_decay": 0.1,
+  "adam_beta1": 0.9,
+  "adam_beta2": 0.95,
+  "adam_epsilon": 1e-08,
+  "optim_target_modules": null,
+  "gradient_accumulation_steps": 4,
+  "average_tokens_across_devices": true,
+  "max_grad_norm": 1.0,
+  "label_smoothing_factor": 0.0,
+  "bf16": true,
+  "fp16": false,
+  "bf16_full_eval": false,
+  "fp16_full_eval": false,
+  "tf32": null,
+  "gradient_checkpointing": true,
+  "gradient_checkpointing_kwargs": null,
+  "torch_compile": false,
+  "torch_compile_backend": null,
+  "torch_compile_mode": null,
+  "use_liger_kernel": false,
+  "liger_kernel_config": null,
+  "use_cache": false,
+  "neftune_noise_alpha": null,
+  "torch_empty_cache_steps": null,
+  "auto_find_batch_size": false,
+  "logging_strategy": "steps",
+  "logging_steps": 5,
+  "logging_first_step": true,
+  "log_on_each_node": true,
+  "logging_nan_inf_filter": true,
+  "include_num_input_tokens_seen": false,
+  "log_level": "passive",
+  "log_level_replica": "warning",
+  "disable_tqdm": null,
+  "report_to": [
+    "tensorboard"
+  ],
+  "run_name": "/home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611",
+  "project": "huggingface",
+  "trackio_space_id": null,
+  "trackio_bucket_id": null,
+  "trackio_static_space_id": null,
+  "eval_strategy": "steps",
+  "eval_steps": 200.0,
+  "eval_delay": 0,
+  "per_device_eval_batch_size": 2,
+  "prediction_loss_only": false,
+  "eval_on_start": false,
+  "eval_do_concat_batches": true,
+  "eval_use_gather_object": false,
+  "eval_accumulation_steps": null,
+  "include_for_metrics": [],
+  "batch_eval_metrics": false,
+  "save_only_model": false,
+  "save_strategy": "steps",
+  "save_steps": 200.0,
+  "save_on_each_node": false,
+  "save_total_limit": 3,
+  "enable_jit_checkpoint": false,
+  "push_to_hub": false,
+  "hub_token": null,
+  "hub_private_repo": null,
+  "hub_model_id": null,
+  "hub_strategy": "every_save",
+  "hub_always_push": false,
+  "hub_revision": null,
+  "load_best_model_at_end": true,
+  "metric_for_best_model": "loss",
+  "greater_is_better": false,
+  "ignore_data_skip": false,
+  "restore_callback_states_from_checkpoint": false,
+  "full_determinism": false,
+  "seed": 42,
+  "data_seed": 42,
+  "use_cpu": false,
+  "accelerator_config": {
+    "dispatch_batches": false
+  },
+  "parallelism_config": null,
+  "dataloader_drop_last": false,
+  "dataloader_num_workers": 2,
+  "dataloader_pin_memory": true,
+  "dataloader_persistent_workers": true,
+  "dataloader_prefetch_factor": null,
+  "dataloader_multiprocessing_context": null,
+  "dataloader_in_order": true,
+  "remove_unused_columns": true,
+  "label_names": null,
+  "train_sampling_strategy": "random",
+  "length_column_name": "length",
+  "ddp_find_unused_parameters": null,
+  "ddp_bucket_cap_mb": null,
+  "ddp_broadcast_buffers": null,
+  "ddp_static_graph": null,
+  "ddp_backend": null,
+  "ddp_timeout": 18000000,
+  "fsdp": [],
+  "fsdp_config": null,
+  "deepspeed": null,
+  "debug": null,
+  "skip_memory_metrics": true,
+  "do_train": false,
+  "do_eval": false,
+  "do_predict": false,
+  "resume_from_checkpoint": null,
+  "local_rank": 0,
+  "sortish_sampler": false,
+  "predict_with_generate": false,
+  "generation_max_length": null,
+  "generation_num_beams": null,
+  "generation_config": null,
+  "tuner_backend": "peft",
+  "vit_gradient_checkpointing": true,
+  "router_aux_loss_coef": 0.0,
+  "enable_dft_loss": false,
+  "enable_channel_loss": false,
+  "safe_serialization": true,
+  "max_shard_size": "5GB",
+  "check_model": true,
+  "acc_strategy": "token",
+  "train_dataloader_shuffle": true,
+  "group_by_length": false,
+  "max_epochs": null,
+  "aligner_lr": null,
+  "vit_lr": 1e-05,
+  "use_logits_to_keep": null,
+  "ds3_gather_for_generation": true,
+  "resume_only_model": false,
+  "optimizer": null,
+  "loss_type": null,
+  "mrl_dims": null,
+  "eval_metric": null,
+  "callbacks": [],
+  "early_stop_interval": null,
+  "eval_use_evalscope": false,
+  "eval_dataset": [],
+  "eval_dataset_args": null,
+  "eval_limit": null,
+  "eval_generation_config": null,
+  "extra_eval_args": null,
+  "tuner_type": "lora",
+  "use_galore": false,
+  "galore_target_modules": null,
+  "galore_rank": 128,
+  "galore_update_proj_gap": 50,
+  "galore_scale": 1.0,
+  "galore_proj_type": "std",
+  "galore_optim_per_parameter": false,
+  "galore_with_embedding": false,
+  "galore_quantization": false,
+  "galore_proj_quant": false,
+  "galore_proj_bits": 4,
+  "galore_proj_group_size": 256,
+  "galore_cos_threshold": 0.4,
+  "galore_gamma_proj": 2,
+  "galore_queue_size": 5,
+  "lisa_activated_layers": 0,
+  "lisa_step_interval": 20,
+  "use_flash_ckpt": false,
+  "logging_dir": "/home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611/runs",
+  "warmup_ratio": 0.05,
+  "use_ray": false,
+  "ray_exp_name": null,
+  "device_groups": null,
+  "model": "/home/jca/models/Qwen3-VL-8B-Instruct",
+  "model_type": "qwen3_vl",
+  "model_revision": null,
+  "task_type": "causal_lm",
+  "torch_dtype": "bfloat16",
+  "attn_impl": null,
+  "experts_impl": null,
+  "new_special_tokens": [],
+  "num_labels": null,
+  "problem_type": null,
+  "rope_scaling": null,
+  "device_map": null,
+  "max_memory": {},
+  "max_model_len": null,
+  "local_repo_path": null,
+  "init_strategy": null,
+  "template": "qwen3_vl",
+  "system": null,
+  "max_length": 4096,
+  "truncation_strategy": "delete",
+  "max_pixels": null,
+  "agent_template": null,
+  "norm_bbox": null,
+  "use_chat_template": true,
+  "padding_side": "right",
+  "padding_free": false,
+  "loss_scale": "default",
+  "sequence_parallel_size": 1,
+  "is_binary_loss_scale": null,
+  "template_backend": "swift",
+  "response_prefix": null,
+  "enable_thinking": null,
+  "preserve_thinking": null,
+  "add_non_thinking_prefix": true,
+  "disable_ignore_empty_think": false,
+  "dataset": [
+    "/home/jca/calligraphy_appreciation/data/sft/train_multitask.jsonl"
+  ],
+  "val_dataset": [
+    "/home/jca/calligraphy_appreciation/data/sft/validation.jsonl"
+  ],
+  "cached_dataset": [],
+  "cached_val_dataset": [],
+  "split_dataset_ratio": 0.0,
+  "dataset_num_proc": 2,
+  "load_from_cache_file": false,
+  "dataset_shuffle": true,
+  "val_dataset_shuffle": false,
+  "streaming": false,
+  "interleave_prob": null,
+  "stopping_strategy": "first_exhausted",
+  "shuffle_buffer_size": 1000,
+  "download_mode": "reuse_dataset_if_exists",
+  "columns": {},
+  "strict": true,
+  "disable_auto_column_mapping": false,
+  "model_name": null,
+  "model_author": null,
+  "custom_dataset_info": [],
+  "decision_drop_instructions": 0.0,
+  "decision_subsample_options": 0,
+  "decision_state_as_text": 0.0,
+  "decision_shuffle_fields": false,
+  "quant_method": "bnb",
+  "quant_bits": 4,
+  "hqq_axis": null,
+  "bnb_4bit_compute_dtype": "bfloat16",
+  "bnb_4bit_quant_type": "nf4",
+  "bnb_4bit_use_double_quant": true,
+  "bnb_4bit_quant_storage": null,
+  "max_new_tokens": 64,
+  "temperature": 0.0,
+  "top_k": null,
+  "top_p": null,
+  "repetition_penalty": null,
+  "num_beams": 1,
+  "stream": false,
+  "stop_words": [],
+  "logprobs": false,
+  "top_logprobs": null,
+  "structured_outputs_regex": null,
+  "adapters": [],
+  "external_plugins": [],
+  "custom_register_path": [],
+  "model_kwargs": {},
+  "enable_npu_model_patch": true,
+  "load_args": false,
+  "load_data_args": false,
+  "packing": false,
+  "packing_length": null,
+  "packing_num_proc": 1,
+  "packing_strategy": "binpack",
+  "lazy_tokenize": true,
+  "use_hf": false,
+  "ignore_args_error": false,
+  "use_swift_lora": false,
+  "freeze_parameters": [],
+  "freeze_parameters_regex": null,
+  "freeze_parameters_ratio": 0.0,
+  "trainable_parameters": [],
+  "trainable_parameters_regex": null,
+  "freeze_llm": false,
+  "freeze_vit": false,
+  "freeze_aligner": true,
+  "target_modules": [
+    "all-linear"
+  ],
+  "target_regex": null,
+  "target_parameters": null,
+  "modules_to_save": [],
+  "lora_rank": 128,
+  "lora_alpha": 256.0,
+  "lora_dropout": 0.05,
+  "lora_bias": "none",
+  "lora_dtype": null,
+  "lorap_lr_ratio": null,
+  "use_rslora": false,
+  "use_dora": false,
+  "lora_ga_batch_size": 2,
+  "lora_ga_iters": 2,
+  "lora_ga_max_length": 1024,
+  "lora_ga_direction": "ArB2r",
+  "lora_ga_scale": "stable",
+  "lora_ga_stable_gamma": 16,
+  "init_weights": true,
+  "fourier_n_frequency": 2000,
+  "fourier_scaling": 300.0,
+  "boft_block_size": 4,
+  "boft_block_num": 0,
+  "boft_n_butterfly_factor": 1,
+  "boft_dropout": 0.0,
+  "vera_rank": 256,
+  "vera_projection_prng_key": 0,
+  "vera_dropout": 0.0,
+  "vera_d_initial": 0.1,
+  "adapter_act": "gelu",
+  "adapter_length": 128,
+  "adalora_target_r": 8,
+  "adalora_init_r": 12,
+  "adalora_tinit": 0,
+  "adalora_tfinal": 0,
+  "adalora_deltaT": 1,
+  "adalora_beta1": 0.85,
+  "adalora_beta2": 0.85,
+  "adalora_orth_reg_weight": 0.5,
+  "llamapro_num_new_blocks": 4,
+  "llamapro_num_groups": null,
+  "reft_layer_key": null,
+  "reft_layers": null,
+  "reft_rank": 4,
+  "reft_intervention_type": "LoreftIntervention",
+  "reft_args": null,
+  "swanlab_token": null,
+  "swanlab_project": "ms-swift",
+  "swanlab_workspace": null,
+  "swanlab_exp_name": null,
+  "swanlab_notification_method": null,
+  "swanlab_webhook_url": null,
+  "swanlab_secret": null,
+  "swanlab_sender_email": null,
+  "swanlab_receiver_email": null,
+  "swanlab_smtp_server": null,
+  "swanlab_smtp_port": null,
+  "swanlab_email_language": "zh",
+  "swanlab_mode": "cloud",
+  "add_version": true,
+  "create_checkpoint_symlink": true,
+  "zero_hpz_partition_size": null,
+  "deepspeed_autotp_size": null,
+  "swift_version": "4.6.0",
+  "ckpt_dir": null,
+  "rank": 0,
+  "global_world_size": 2,
+  "local_world_size": 2,
+  "model_suffix": "Qwen3-VL-8B-Instruct",
+  "model_info": "ModelInfo(model_type='qwen3_vl', model_dir='/home/jca/models/Qwen3-VL-8B-Instruct', torch_dtype=torch.bfloat16, max_model_len=262144, quant_method='bnb', quant_bits=4, rope_scaling=None, is_moe_model=False, is_multimodal=True, config=None, task_type='causal_lm', num_labels=None)",
+  "model_meta": "ModelMeta(model_type='qwen3_vl', model_groups=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen3-VL-2B-Instruct', hf_model_id='Qwen/Qwen3-VL-2B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-2B-Thinking', hf_model_id='Qwen/Qwen3-VL-2B-Thinking', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-2B-Instruct-FP8', hf_model_id='Qwen/Qwen3-VL-2B-Instruct-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-2B-Thinking-FP8', hf_model_id='Qwen/Qwen3-VL-2B-Thinking-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-4B-Instruct', hf_model_id='Qwen/Qwen3-VL-4B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-4B-Thinking', hf_model_id='Qwen/Qwen3-VL-4B-Thinking', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-4B-Instruct-FP8', hf_model_id='Qwen/Qwen3-VL-4B-Instruct-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-4B-Thinking-FP8', hf_model_id='Qwen/Qwen3-VL-4B-Thinking-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-8B-Instruct', hf_model_id='Qwen/Qwen3-VL-8B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-8B-Thinking', hf_model_id='Qwen/Qwen3-VL-8B-Thinking', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-8B-Instruct-FP8', hf_model_id='Qwen/Qwen3-VL-8B-Instruct-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-8B-Thinking-FP8', hf_model_id='Qwen/Qwen3-VL-8B-Thinking-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-32B-Instruct', hf_model_id='Qwen/Qwen3-VL-32B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-32B-Thinking', hf_model_id='Qwen/Qwen3-VL-32B-Thinking', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-32B-Instruct-FP8', hf_model_id='Qwen/Qwen3-VL-32B-Instruct-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-32B-Thinking-FP8', hf_model_id='Qwen/Qwen3-VL-32B-Thinking-FP8', model_path=None, ms_revision=None, hf_revision=None)], template='qwen3_vl', ignore_patterns=None, requires=None, tags=[])], loader=<class 'swift.model.models.qwen.Qwen3VLLoader'>, template='qwen3_vl', model_arch=MultiModelKeys(arch_name='qwen3_vl', embedding=None, module_list=None, lm_head=None, q_proj=None, k_proj=None, v_proj=None, o_proj=None, attention=None, mlp='model.language_model.layers.{}.mlp', down_proj=None, qkv_proj=None, qk_proj=None, qa_proj=None, qb_proj=None, kv_proj=None, kva_proj=None, kvb_proj=None, language_model=['model.language_model', 'lm_head'], aligner=['model.visual.merger', 'model.visual.deepstack_merger_list'], vision_tower=['model.visual'], generator=[]), mcore_model_type=None, architectures=['Qwen3VLForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=['transformers>=4.57', 'qwen_vl_utils>=0.0.14', 'decord'], tags=[])",
+  "model_dir": "/home/jca/models/Qwen3-VL-8B-Instruct",
+  "template_meta": "QwenTemplateMeta(template_type='qwen3_vl', prefix=[], prompt=['<|im_start|>user\\n{{QUERY}}<|im_end|>\\n<|im_start|>assistant\\n'], chat_sep=['<|im_end|>\\n'], suffix=['<|im_end|>\\n'], template_cls=<class 'swift.template.templates.qwen.Qwen3VLTemplate'>, system_prefix=['<|im_start|>system\\n{{SYSTEM}}<|im_end|>\\n'], default_system=None, auto_add_bos=False, stop_words=['<|endoftext|>'], agent_template='hermes', is_thinking=False, thinking_prefix='<think>\\n', non_thinking_prefix='', preserve_thinking=None, history_thinking_prefix='')",
+  "_val_dataset_exists": true,
+  "hub": "<class 'swift.hub.hub.MSHub'>",
+  "evaluation_strategy": "steps",
+  "training_args": "Seq2SeqTrainingArguments(output_dir='/home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611', per_device_train_batch_size=2, num_train_epochs=4.0, max_steps=-1, learning_rate=0.0001, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_steps=0.05, optim=<OptimizerNames.ADAMW_TORCH_FUSED: 'adamw_torch_fused'>, optim_args=None, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, optim_target_modules=None, gradient_accumulation_steps=4, average_tokens_across_devices=None, max_grad_norm=1.0, label_smoothing_factor=0.0, bf16=True, fp16=False, bf16_full_eval=False, fp16_full_eval=False, tf32=None, gradient_checkpointing=True, gradient_checkpointing_kwargs=None, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, use_liger_kernel=False, liger_kernel_config=None, use_cache=False, neftune_noise_alpha=None, torch_empty_cache_steps=None, auto_find_batch_size=False, logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_steps=5, logging_first_step=True, log_on_each_node=True, logging_nan_inf_filter=True, include_num_input_tokens_seen=None, log_level='passive', log_level_replica='warning', disable_tqdm=False, report_to=['tensorboard'], run_name='/home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611', project='huggingface', trackio_space_id=None, trackio_bucket_id=None, trackio_static_space_id=None, eval_strategy=<IntervalStrategy.STEPS: 'steps'>, eval_steps=200, eval_delay=0, per_device_eval_batch_size=2, prediction_loss_only=False, eval_on_start=False, eval_do_concat_batches=True, eval_use_gather_object=False, eval_accumulation_steps=None, include_for_metrics=[], batch_eval_metrics=False, save_only_model=False, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=200, save_on_each_node=False, save_total_limit=3, enable_jit_checkpoint=False, push_to_hub=False, hub_token=None, hub_private_repo=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_always_push=False, hub_revision=None, load_best_model_at_end=True, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, restore_callback_states_from_checkpoint=False, full_determinism=False, seed=42, data_seed=42, use_cpu=False, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), parallelism_config=None, dataloader_drop_last=False, dataloader_num_workers=2, dataloader_pin_memory=True, dataloader_persistent_workers=True, dataloader_prefetch_factor=2, dataloader_multiprocessing_context=None, dataloader_in_order=True, remove_unused_columns=False, label_names=None, train_sampling_strategy='random', length_column_name='length', ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, ddp_static_graph=None, ddp_backend=None, ddp_timeout=18000000, fsdp=[], fsdp_config=None, deepspeed=None, debug=[], skip_memory_metrics=True, do_train=False, do_eval=True, do_predict=False, resume_from_checkpoint=None, local_rank=0, sortish_sampler=False, predict_with_generate=False, generation_max_length=None, generation_num_beams=None, generation_config=None, tuner_backend='peft', vit_gradient_checkpointing=True, router_aux_loss_coef=0.0, enable_dft_loss=False, enable_channel_loss=False, safe_serialization=True, max_shard_size='5GB', check_model=True, acc_strategy='token', train_dataloader_shuffle=True, group_by_length=False, max_epochs=None, aligner_lr=None, vit_lr=1e-05, use_logits_to_keep=None, ds3_gather_for_generation=True, resume_only_model=False, optimizer='multimodal', loss_type=None, mrl_dims=None, eval_metric=None, callbacks=[], early_stop_interval=None, eval_use_evalscope=False, eval_dataset=[], eval_dataset_args=None, eval_limit=None, eval_generation_config=None, extra_eval_args=None, tuner_type='lora', use_galore=False, galore_target_modules=None, galore_rank=128, galore_update_proj_gap=50, galore_scale=1.0, galore_proj_type='std', galore_optim_per_parameter=False, galore_with_embedding=False, galore_quantization=False, galore_proj_quant=False, galore_proj_bits=4, galore_proj_group_size=256, galore_cos_threshold=0.4, galore_gamma_proj=2, galore_queue_size=5, lisa_activated_layers=0, lisa_step_interval=20, use_flash_ckpt=False, logging_dir='/home/jca/models/trained_models/sft/qwen3vl_sft_v16/v0-20261028-121611/runs', warmup_ratio=0.0)"
+}

sft/qwen3vl_sft_v16/v0-20261028-121611/checkpoint-5000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f77a58860a563c955e0d1c872dd4deced44b2bcd6f9a414cc3a7f3f0ef4e1672
+size 3286824211