internal/

directory
v0.3.1 Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: Aug 7, 2026 License: AGPL-3.0

Directories

Path Synopsis
Package cli defines the hay command-line interface.
Package cli defines the hay command-line interface.
Package daemon is hay's long-running process: a localhost REST API and a filesystem watcher that keeps the index fresh.
Package daemon is hay's long-running process: a localhost REST API and a filesystem watcher that keeps the index fresh.
Package embed turns text into vectors for semantic search.
Package embed turns text into vectors for semantic search.
bundled
Package bundled is the in-binary embedding backend: a pure-Go implementation of all-MiniLM-L6-v2 (WordPiece tokenizer + 6-layer BERT encoder + mean pooling).
Package bundled is the in-binary embedding backend: a pure-Go implementation of all-MiniLM-L6-v2 (WordPiece tokenizer + 6-layer BERT encoder + mean pooling).
bundled/quantize command
Command quantize converts fp32 safetensors weights to weight-only int8: every large 2-D tensor becomes an I8 tensor plus a per-row F32 "<name>.scale" companion.
Command quantize converts fp32 safetensors weights to weight-only int8: every large 2-D tensor becomes an I8 tensor plus a per-row F32 "<name>.scale" companion.
Package index owns Haypile's storage: a single SQLite file holding sources, files, chunks, and the FTS5 keyword index.
Package index owns Haypile's storage: a single SQLite file holding sources, files, chunks, and the FTS5 keyword index.
Package ingest turns documents on disk into indexed chunks: it walks folders, extracts text with provenance (page numbers, headings), and splits it for the index.
Package ingest turns documents on disk into indexed chunks: it walks folders, extracts text with provenance (page numbers, headings), and splits it for the index.
Package llm speaks to whatever OpenAI-compatible server the user already runs (Ollama, LM Studio, llama.cpp, Jan, …) — or, when they bring their own API key, whatever cloud endpoint they point it at.
Package llm speaks to whatever OpenAI-compatible server the user already runs (Ollama, LM Studio, llama.cpp, Jan, …) — or, when they bring their own API key, whatever cloud endpoint they point it at.
Package query runs retrieval: keyword search, vector search, and the hybrid merge of the two.
Package query runs retrieval: keyword search, vector search, and the hybrid merge of the two.
Package webui is the bundled single-user web interface, served by the daemon at /.
Package webui is the bundled single-user web interface, served by the daemon at /.

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL