[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78T5FJJS8X58G3JG0JS9898":3,"skill-same:01M78T5FJJS8X58G3JG0JS9898":88},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":40,"commitSha":41,"skillDirectory":42,"forksCount":43,"repositoryHtmlUrl":44,"repositoryHomepageUrl":6,"ownerHtmlUrl":45,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":46,"install":47,"resources":59},"01M78T5FJJS8X58G3JG0JS9898","inference",null,"Operate Baichuan2 chat\u002Fbase inference, terminal chat, Streamlit chat, and OpenAI-compatible chat-completions workflows.","VectorSpaceLab\u002FAREX-Skill","VectorSpaceLab","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F180637102?v=4","01M7538C5WPSJGWF7KHR4EFZ9F","01M7538C6R2GFP53NPXTCZNBRT",351,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fbaichuan2\u002Fsub-skills\u002Finference\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: inference\ndescription: \"Operate Baichuan2 chat\u002Fbase inference, terminal chat, Streamlit\n  chat, and OpenAI-compatible chat-completions workflows.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\n  repo: Baichuan2\n  sub-skill-id: inference\n  supported-workflows:\n    - python-chat-inference\n    - python-base-generation\n    - interactive-cli-chat\n    - streamlit-web-chat\n    - openai-compatible-chat-completions\n  required-backend: cuda-for-primary-chat-demos\n  verified-inspection-context:\n    torch: 2.5.1+cu121\n    transformers: 5.15.0\n    cuda-smoke: passed-on-nvidia-a100\n    streamlit-cli-help: passed\n    deepspeed-cli-help: passed\nlicense: Apache 2.0\n---\n\n# Baichuan2 inference\n\nUse this sub-skill when the user wants to run or understand Baichuan2 inference: Python chat snippets, Base-model text generation, the interactive terminal demo, the Streamlit chat UI, or the OpenAI-compatible `\u002Fv1\u002Fchat\u002Fcompletions` server.\n\nDo **not** route quantization, CPU deployment conversion, model checkpoint normalization, fine-tuning, DeepSpeed training, or LoRA work here. Send those requests to the deployment or fine-tuning sub-skill.\n\n## Route by user intent\n\n| User asks for | Route | Key files |\n| --- | --- | --- |\n| \"chat with Baichuan2 in Python\" or \"use model.chat\" | Chat Python recipe | [`references\u002Fworkflows.md`](references\u002Fworkflows.md#chat-model-python-inference) |\n| \"run the Base model\" or \"plain text completion\" | Base-model generation recipe | [`references\u002Fworkflows.md`](references\u002Fworkflows.md#base-model-python-generation) |\n| \"interactive CLI\" or \"terminal demo\" | Chat CLI helper | [`scripts\u002Fchat_cli.py`](scripts\u002Fchat_cli.py) |\n| \"web demo\" or \"browser UI\" | Streamlit helper | [`scripts\u002Fchat_web_demo.py`](scripts\u002Fchat_web_demo.py) |\n| \"OpenAI API compatible server\" or \"chat completions endpoint\" | Flask API helper | [`scripts\u002Frun_openai_api.py`](scripts\u002Frun_openai_api.py), [`references\u002Fapi-reference.md`](references\u002Fapi-reference.md) |\n| Load\u002Fmodel\u002Fruntime failure | Troubleshooting | [`references\u002Ftroubleshooting.md`](references\u002Ftroubleshooting.md) |\n\n## Core operating facts\n\n- Baichuan2 publishes **Chat** and **Base** model families for 7B and 13B. Use `*-Chat` for assistant-style conversations and `*-Base` for continuation\u002Fgeneration tasks.\n- Chat-model code uses `AutoTokenizer.from_pretrained(..., use_fast=False, trust_remote_code=True)`, `AutoModelForCausalLM.from_pretrained(..., device_map=\"auto\", torch_dtype=...)`, and `GenerationConfig.from_pretrained(model_id)` before calling `model.chat(tokenizer, messages)`.\n- Base-model code uses tokenization plus `model.generate(...)`; do not use the CLI\u002Fweb\u002FAPI wrappers with Base checkpoints because those wrappers expect Chat model behavior.\n- `device_map=\"auto\"` lets Transformers place weights across available GPUs. To constrain visible GPUs, set `CUDA_VISIBLE_DEVICES` before launch.\n- `torch.bfloat16` is the README chat-inference dtype and is appropriate on Ampere-class GPUs such as A100. The bundled chat\u002FAPI demos default to `float16` to match the native demo scripts; use `--dtype bfloat16` on hardware that supports it.\n- The prepared inspection environment verified a CUDA-capable stack with torch `2.5.1+cu121`, transformers `5.15.0`, Streamlit CLI help, DeepSpeed CLI help, and a CUDA tensor smoke test on NVIDIA A100. DeepSpeed is not required for this inference sub-skill.\n\n## Minimal chat model recipe\n\n```python\nimport torch\nfrom transformers import AutoModelForCausalLM, AutoTokenizer\nfrom transformers.generation.utils import GenerationConfig\n\nmodel_id = \"baichuan-inc\u002FBaichuan2-13B-Chat\"\ntokenizer = AutoTokenizer.from_pretrained(\n    model_id,\n    use_fast=False,\n    trust_remote_code=True,\n)\nmodel = AutoModelForCausalLM.from_pretrained(\n    model_id,\n    device_map=\"auto\",\n    torch_dtype=torch.bfloat16,\n    trust_remote_code=True,\n)\nmodel.generation_config = GenerationConfig.from_pretrained(model_id)\n\nmessages = [{\"role\": \"user\", \"content\": \"解释一下“温故而知新”\"}]\nresponse = model.chat(tokenizer, messages)\nprint(response)\n```\n\nFor streaming chat, call `model.chat(tokenizer, messages, stream=True)`. The iterator yields progressively longer response strings, so terminal UIs should print only the delta after the previous position.\n\n## Bundled helpers\n\nRun `--help` before loading weights:\n\n```bash\npython scripts\u002Fchat_cli.py --help\npython scripts\u002Frun_openai_api.py --help\npython scripts\u002Fchat_web_demo.py --help\n```\n\nCheck launch configuration without downloading\u002Floading weights:\n\n```bash\npython scripts\u002Fchat_cli.py --dry-run --model baichuan-inc\u002FBaichuan2-13B-Chat\npython scripts\u002Frun_openai_api.py --dry-run --host 127.0.0.1 --port 8000\npython scripts\u002Fchat_web_demo.py --dry-run --model baichuan-inc\u002FBaichuan2-13B-Chat\n```\n\nLaunch examples:\n\n```bash\n# Terminal chat, Chat model only.\npython scripts\u002Fchat_cli.py --model baichuan-inc\u002FBaichuan2-13B-Chat --dtype float16 --stream\n\n# OpenAI-compatible non-streaming API.\npython scripts\u002Frun_openai_api.py --model baichuan-inc\u002FBaichuan2-13B-Chat --host 0.0.0.0 --port 8000 --dtype float16\n\n# Streamlit web UI. Streamlit server flags configure the web host\u002Fport;\n# arguments after `--` are passed to the Baichuan2 helper.\nstreamlit run scripts\u002Fchat_web_demo.py --server.address 0.0.0.0 --server.port 8501 -- --model baichuan-inc\u002FBaichuan2-13B-Chat --dtype float16\n```\n\n## API behavior to remember\n\n- Endpoint: `POST \u002Fv1\u002Fchat\u002Fcompletions`.\n- Request body should contain `messages`, each with `role` and `content`.\n- The bundled API helper is intentionally **non-streaming**. If a client sends `\"stream\": true`, it returns HTTP 400 with a streaming-not-supported error.\n- The API helper loads one configured Chat model at process startup. The request `model` field is accepted for OpenAI-style clients but does not switch the already-loaded model.\n\n## When to escalate\n\n- Missing weights, private model access, incompatible remote code, or network timeouts: use [`references\u002Ftroubleshooting.md`](references\u002Ftroubleshooting.md#model-weights-network-and-cache).\n- Out-of-memory or dtype failures: use [`references\u002Ftroubleshooting.md`](references\u002Ftroubleshooting.md#gpu-memory-pressure-and-dtype-choice).\n- Need 4-bit\u002F8-bit loading, CPU-only operation, or checkpoint conversion: route to deployment.\n- Need SFT, LoRA, DeepSpeed, training data validation, or model output checkpoints: route to fine-tuning.\n","# Baichuan2 inference\n\nUse this sub-skill when the user wants to run or understand Baichuan2 inference: Python chat snippets, Base-model text generation, the interactive terminal demo, the Streamlit chat UI, or the OpenAI-compatible `\u002Fv1\u002Fchat\u002Fcompletions` server.\n\nDo **not** route quantization, CPU deployment conversion, model checkpoint normalization, fine-tuning, DeepSpeed training, or LoRA work here. Send those requests to the deployment or fine-tuning sub-skill.\n\n## Route by user intent\n\n| User asks for | Route | Key files |\n| --- | --- | --- |\n| \"chat with Baichuan2 in Python\" or \"use model.chat\" | Chat Python recipe | [`references\u002Fworkflows.md`](references\u002Fworkflows.md#chat-model-python-inference) |\n| \"run the Base model\" or \"plain text completion\" | Base-model generation recipe | [`references\u002Fworkflows.md`](references\u002Fworkflows.md#base-model-python-generation) |\n| \"interactive CLI\" or \"terminal demo\" | Chat CLI helper | [`scripts\u002Fchat_cli.py`](scripts\u002Fchat_cli.py) |\n| \"web demo\" or \"browser UI\" | Streamlit helper | [`scripts\u002Fchat_web_demo.py`](scripts\u002Fchat_web_demo.py) |\n| \"OpenAI API compatible server\" or \"chat completions endpoint\" | Flask API helper | [`scripts\u002Frun_openai_api.py`](scripts\u002Frun_openai_api.py), [`references\u002Fapi-reference.md`](references\u002Fapi-reference.md) |\n| Load\u002Fmodel\u002Fruntime failure | Troubleshooting | [`references\u002Ftroubleshooting.md`](references\u002Ftroubleshooting.md) |\n\n## Core operating facts\n\n- Baichuan2 publishes **Chat** and **Base** model families for 7B and 13B. Use `*-Chat` for assistant-style conversations and `*-Base` for continuation\u002Fgeneration tasks.\n- Chat-model code uses `AutoTokenizer.from_pretrained(..., use_fast=False, trust_remote_code=True)`, `AutoModelForCausalLM.from_pretrained(..., device_map=\"auto\", torch_dtype=...)`, and `GenerationConfig.from_pretrained(model_id)` before calling `model.chat(tokenizer, messages)`.\n- Base-model code uses tokenization plus `model.generate(...)`; do not use the CLI\u002Fweb\u002FAPI wrappers with Base checkpoints because those wrappers expect Chat model behavior.\n- `device_map=\"auto\"` lets Transformers place weights across available GPUs. To constrain visible GPUs, set `CUDA_VISIBLE_DEVICES` before launch.\n- `torch.bfloat16` is the README chat-inference dtype and is appropriate on Ampere-class GPUs such as A100. The bundled chat\u002FAPI demos default to `float16` to match the native demo scripts; use `--dtype bfloat16` on hardware that supports it.\n- The prepared inspection environment verified a CUDA-capable stack with torch `2.5.1+cu121`, transformers `5.15.0`, Streamlit CLI help, DeepSpeed CLI help, and a CUDA tensor smoke test on NVIDIA A100. DeepSpeed is not required for this inference sub-skill.\n\n## Minimal chat model recipe\n\n```python\nimport torch\nfrom transformers import AutoModelForCausalLM, AutoTokenizer\nfrom transformers.generation.utils import GenerationConfig\n\nmodel_id = \"baichuan-inc\u002FBaichuan2-13B-Chat\"\ntokenizer = AutoTokenizer.from_pretrained(\n    model_id,\n    use_fast=False,\n    trust_remote_code=True,\n)\nmodel = AutoModelForCausalLM.from_pretrained(\n    model_id,\n    device_map=\"auto\",\n    torch_dtype=torch.bfloat16,\n    trust_remote_code=True,\n)\nmodel.generation_config = GenerationConfig.from_pretrained(model_id)\n\nmessages = [{\"role\": \"user\", \"content\": \"解释一下“温故而知新”\"}]\nresponse = model.chat(tokenizer, messages)\nprint(response)\n```\n\nFor streaming chat, call `model.chat(tokenizer, messages, stream=True)`. The iterator yields progressively longer response strings, so terminal UIs should print only the delta after the previous position.\n\n## Bundled helpers\n\nRun `--help` before loading weights:\n\n```bash\npython scripts\u002Fchat_cli.py --help\npython scripts\u002Frun_openai_api.py --help\npython scripts\u002Fchat_web_demo.py --help\n```\n\nCheck launch configuration without downloading\u002Floading weights:\n\n```bash\npython scripts\u002Fchat_cli.py --dry-run --model baichuan-inc\u002FBaichuan2-13B-Chat\npython scripts\u002Frun_openai_api.py --dry-run --host 127.0.0.1 --port 8000\npython scripts\u002Fchat_web_demo.py --dry-run --model baichuan-inc\u002FBaichuan2-13B-Chat\n```\n\nLaunch examples:\n\n```bash\n# Terminal chat, Chat model only.\npython scripts\u002Fchat_cli.py --model baichuan-inc\u002FBaichuan2-13B-Chat --dtype float16 --stream\n\n# OpenAI-compatible non-streaming API.\npython scripts\u002Frun_openai_api.py --model baichuan-inc\u002FBaichuan2-13B-Chat --host 0.0.0.0 --port 8000 --dtype float16\n\n# Streamlit web UI. Streamlit server flags configure the web host\u002Fport;\n# arguments after `--` are passed to the Baichuan2 helper.\nstreamlit run scripts\u002Fchat_web_demo.py --server.address 0.0.0.0 --server.port 8501 -- --model baichuan-inc\u002FBaichuan2-13B-Chat --dtype float16\n```\n\n## API behavior to remember\n\n- Endpoint: `POST \u002Fv1\u002Fchat\u002Fcompletions`.\n- Request body should contain `messages`, each with `role` and `content`.\n- The bundled API helper is intentionally **non-streaming**. If a client sends `\"stream\": true`, it returns HTTP 400 with a streaming-not-supported error.\n- The API helper loads one configured Chat model at process startup. The request `model` field is accepted for OpenAI-style clients but does not switch the already-loaded model.\n\n## When to escalate\n\n- Missing weights, private model access, incompatible remote code, or network timeouts: use [`references\u002Ftroubleshooting.md`](references\u002Ftroubleshooting.md#model-weights-network-and-cache).\n- Out-of-memory or dtype failures: use [`references\u002Ftroubleshooting.md`](references\u002Ftroubleshooting.md#gpu-memory-pressure-and-dtype-choice).\n- Need 4-bit\u002F8-bit loading, CPU-only operation, or checkpoint conversion: route to deployment.\n- Need SFT, LoRA, DeepSpeed, training data validation, or model output checkpoints: route to fine-tuning.\n",{"name":5,"license":23,"metadata":24,"description":7,"disable-model-invocation":39},"Apache 2.0",{"repo":25,"disco-role":26,"sub-skill-id":5,"required-backend":27,"supported-workflows":28,"verified-inspection-context":34},"Baichuan2","operating","cuda-for-primary-chat-demos",[29,30,31,32,33],"python-chat-inference","python-base-generation","interactive-cli-chat","streamlit-web-chat","openai-compatible-chat-completions",{"torch":35,"cuda-smoke":36,"transformers":37,"deepspeed-cli-help":38,"streamlit-cli-help":38},"2.5.1+cu121","passed-on-nvidia-a100","5.15.0","passed",true,"main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","skills\u002Frepositories\u002Frepo-skills\u002Fbaichuan2\u002Fsub-skills\u002Finference",27,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab","2026-09-03 10:25:21",{"source":8,"skillName":5,"commands":48,"prompt":58,"zipAvailable":39},[49,52,55],{"tool":50,"command":51},"npx","npx skills add VectorSpaceLab\u002FAREX-Skill --skill inference",{"tool":53,"command":54},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill inference",{"tool":56,"command":57},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill inference","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fbaichuan2\u002Fsub-skills\u002Finference\u002FSKILL.md",[60,66,70,74,80,84],{"path":61,"resourceType":62,"resourceTypeDesc":63,"sizeBytes":64,"sourceHtmlUrl":65},"references\u002Fapi-reference.md","REFERENCE","参考文档",4231,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fbaichuan2\u002Fsub-skills\u002Finference\u002Freferences\u002Fapi-reference.md",{"path":67,"resourceType":62,"resourceTypeDesc":63,"sizeBytes":68,"sourceHtmlUrl":69},"references\u002Ftroubleshooting.md",6305,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fbaichuan2\u002Fsub-skills\u002Finference\u002Freferences\u002Ftroubleshooting.md",{"path":71,"resourceType":62,"resourceTypeDesc":63,"sizeBytes":72,"sourceHtmlUrl":73},"references\u002Fworkflows.md",6842,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fbaichuan2\u002Fsub-skills\u002Finference\u002Freferences\u002Fworkflows.md",{"path":75,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":78,"sourceHtmlUrl":79},"scripts\u002Fchat_cli.py","SCRIPT","脚本",10185,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fbaichuan2\u002Fsub-skills\u002Finference\u002Fscripts\u002Fchat_cli.py",{"path":81,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":82,"sourceHtmlUrl":83},"scripts\u002Fchat_web_demo.py",8288,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fbaichuan2\u002Fsub-skills\u002Finference\u002Fscripts\u002Fchat_web_demo.py",{"path":85,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":86,"sourceHtmlUrl":87},"scripts\u002Frun_openai_api.py",8387,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fbaichuan2\u002Fsub-skills\u002Finference\u002Fscripts\u002Frun_openai_api.py",[89,96,103,110,117,124],{"skillNo":90,"name":91,"nameZh":6,"description":92,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":93,"categories":94,"keywords":95,"publishedAt":18},"01M799NYSH1R3RVAHSM2D88Y84","passnet-triton-opt","Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block\u002Fgrid\u002Fwarp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-triton-opt\u002FSKILL.md",[],[],{"skillNo":97,"name":98,"nameZh":6,"description":99,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":100,"categories":101,"keywords":102,"publishedAt":18},"01M799NS5APEC0N1FZ7XZVW6BN","passnet-solve","END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-solve\u002FSKILL.md",[],[],{"skillNo":104,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":107,"categories":108,"keywords":109,"publishedAt":18},"01M799NNY9BKSM21V5R7QVTBJX","passnet-skill","PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under .\u002Fpass_dir\u002F, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-skill\u002FSKILL.md",[],[],{"skillNo":111,"name":112,"nameZh":6,"description":113,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":114,"categories":115,"keywords":116,"publishedAt":18},"01M799NHH6NP7Z1RK77KGXAN6S","passnet-pattern-fusion","Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()\u002Freplacement_args()\u002Freplacement_func() files, when a pass \"failed to match\", or when deciding how to fuse multiple kernels\u002Fops.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-pattern-fusion\u002FSKILL.md",[],[],{"skillNo":118,"name":119,"nameZh":6,"description":120,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":121,"categories":122,"keywords":123,"publishedAt":18},"01M799ND2J4WNC2PGAZ0A06NDF","passnet-orchestrate","Round\u002Fbatch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep\u002Frevert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-orchestrate\u002FSKILL.md",[],[],{"skillNo":125,"name":126,"nameZh":6,"description":127,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":128,"categories":129,"keywords":130,"publishedAt":18},"01M799N8SBTMY02ZDG2AZF5ZWE","passnet-feedback","Fast, deterministic feedback for PassNet work: pre-flight pattern\u002Fmatch verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-feedback\u002FSKILL.md",[],[]]