embeddings-and-generative
Select Chonkie embedding models, provider embeddings, and generative chunking workflows without hidden network assumptions.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- embeddings-and-generative
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Select Chonkie embedding models, provider embeddings, and generative chunking workflows without hidden network assumptions.
- disable-model-invocation
- true
02 资源文件 4
- 参考文档references/embeddings-reference.md12.6 KB
- 参考文档references/model-dependent-chunking.md10.9 KB
- 参考文档references/troubleshooting.md10.8 KB
- 脚本scripts/optional_dependency_probe.py9.7 KB
03 SKILL.md 正文
Chonkie embeddings and generative workflows
Use this sub-skill when a task depends on embedding vectors, semantic/late/neural/LLM chunking, embedding-based refinement, or provider-backed generation inside Chonkie.
Route here for
- Selecting or wiring
AutoEmbeddings,BaseEmbeddings,Model2VecEmbeddings,SentenceTransformerEmbeddings,CatsuEmbeddings, provider embedding wrappers, orLiteLLMEmbeddings. - Adding chunk embeddings with
EmbeddingsRefinerybefore retrieval, export, or vector storage. - Operating
SemanticChunker,LateChunker,NeuralChunker, orSlumberChunkerwith explicit optional dependencies, model cache/download, credential, batching, and timeout decisions. - Debugging provider
Genieclasses used bySlumberChunker: OpenAI, Azure OpenAI, Gemini, Groq, and Cerebras.
Route elsewhere first
- Deterministic token/sentence/recursive/table/code chunking and
Chunk/tokenizer contracts:../chunking-and-types/. - Pipeline placement, CHOMP ordering, file processing, chefs, and refinery orchestration:
../pipelines-and-processing/. - Chonkie Cloud, local FastAPI/API schemas, and CLI deployment:
../interfaces-and-deployment/. - Vector DB handshakes and chunk/document storage targets:
../integrations-and-storage/.
Operating procedure
- Identify whether the user needs local embeddings, third-party provider embeddings, model-dependent chunking, or generative LLM chunking.
- Read
references/embeddings-reference.mdfor embedding/provider selection, constructor arguments, extras, credentials, and batching/timeout notes. - Read
references/model-dependent-chunking.mdbefore usingEmbeddingsRefinery,SemanticChunker,LateChunker,NeuralChunker, orSlumberChunker. - Run
scripts/optional_dependency_probe.pyto inspect installed optional modules without making network calls. Use--instantiate-safeonly for the script's safe no-network checks. - If optional dependencies, credentials, models, or services are missing, read
references/troubleshooting.mdand select a deterministic fallback or ask the user to authorize the required install/network/API-key path.
Verification boundary
The Chonkie skill's required environment only verified CPU-local core package behavior. Model downloads, accelerator use, third-party provider APIs, provider credentials, and live generative calls are optional and must be treated as unverified unless the current user explicitly supplies and authorizes those resources.
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp 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.
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.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
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()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert 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.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match 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).