guanghulab/brain/training-status.json

77 lines
3.2 KiB
JSON
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

{
"version": "1.3",
"last_updated": "2026-05-20T09:48+08:00",
"updated_by": "铸渊 (ICE-GL-ZY001)",
"mother_model": {
"status": "completed",
"status_label": "✅ 已完成 · 已上传COS bingshuo-1317346199",
"model": "Qwen/Qwen2.5-7B",
"final_loss": 1.5229,
"final_epoch": 3,
"total_epochs": 3,
"cos_path": "cos://bingshuo-1317346199/models/qwen25-7b-sft/final/",
"completed_at": "2026-05-19T~22:30+08:00",
"fixes": {
"generation_config_eos": "151645 ✅",
"default_system": "已清除 ✅"
}
},
"code_model": {
"status": "completed",
"status_label": "✅ 已完成 · 已上传COS bingshuo-1317346199",
"model": "Qwen/Qwen2.5-Coder-7B",
"final_loss": 1.822,
"cos_path": "cos://bingshuo-1317346199/models/qwen25-coder-7b-sft/final/",
"fixes": {}
},
"distill_shuangyan_15b": {
"status": "running",
"status_label": "🔴 蒸馏训练中 · Ep1/3 step~4850/15781 loss≈1.96",
"version": "v7",
"script": "scripts/distill_mother_v7.py",
"teacher": "母模型7B SFT (COS bingshuo)",
"student": "Qwen2.5-1.5B-Instruct",
"params": "B=2, GA=16, LR=1e-5, TEMP=2.0, ALPHA=0.7",
"host": "connect.westd.seetacloud.com:23647 (RTX PRO 6000 96GB)",
"started_at": "2026-05-20T09:23+08:00",
"estimated_completion": "~12:00~13:00+08:00 (3 epoch, ~45min/ep)",
"cos_output": "models/shuangyan-15b-distill/",
"pipeline_watchdog": "/tmp/distill_pipeline.sh (PID 581561, 自动触发Track2)"
},
"distill_coder_15b": {
"status": "pending",
"status_label": "⏳ 等待Track1完成 + 管道自动触发",
"script": "scripts/distill_coder_v2.py",
"teacher": "代码模型7B SFT (COS bingshuo)",
"student": "Qwen2.5-Coder-1.5B-Instruct",
"params": "B=2, GA=16, LR=1e-5 (预修OOM)",
"cos_output": "models/qwen25-15b-coder-distill/"
},
"pipeline": {
"steps": [
{"name": "母模型SFT训练+上传COS", "status": "completed", "label": "✅"},
{"name": "代码模型SFT训练+上传COS", "status": "completed", "label": "✅"},
{"name": "Track1 母模型→霜砚1.5B蒸馏", "status": "running", "label": "🔄 Ep1/3 step~4850/15781"},
{"name": "Track1 上传COS", "status": "pending", "label": "⏳ 蒸馏自带上传"},
{"name": "Track2 代码模型→铸渊1.5B蒸馏", "status": "pending", "label": "⏳ 管道自动触发"},
{"name": "Track2 上传COS", "status": "pending", "label": "⏳"}
]
},
"website_status": {
"login_api": "✅ 正常",
"training_panel": "✅ 实时显示 (fetch_train.py v4)",
"js_all_functions": "✅ 已修复 (setInterval恢复)",
"gpu_monitoring": "✅ training-status.json每分钟更新"
},
"alerts": [
"⚠️ 本次修复了fetch_train.py v3的3个bugphase检测/epoch提取/tail长度",
"⚠️ 前一次index.html修复(3583413)引入语法错误→全站JS崩溃→已修复",
"⚠️ Track2的OOM已在Track1完成前预修",
"⚠️ 冰朔强调的工作流程:先读仓库记忆→理解全局→再动手"
],
"links": {
"notion": "https://www.notion.so/363fb92f383181029fafef8438d6c300",
"repo": "https://guanghulab.com/code/bingshuo/guanghulab"
}
}