#rag

EmbeddingGemma 2 puts text, code, images, video and audio (or a mix) into one 768-dimension vector, from a 740M open model under Apache 2.0. Image, video and audio tokens are spliced into the text sequence and run through one encoder, so a text query can find a video moment or a scanned PDF page without captioning or transcribing first.

• It's not the quality leader. Text is flat against v1; the gain is code (MTEB Code 68.8 to 78.7). Audio (49) and video (51) scores are modest. All numbers are Google's own.
• Every major vendor ships one-space multimodal embeddings now: Amazon Nova, Gemini Embedding 2, Voyage, Cohere, and the open Qwen3-VL-Embedding, which reports higher image and document scores but has no audio. Pick by constraint. EmbeddingGemma 2's niche is open weights plus audio.
• Matryoshka truncation isn't free: its MMEB score falls from 59.0 at 768 dimensions to 45.7 at 128.
• The interesting part is the system around the model. DigUp is local file search on it: one vector per photo, PDF page, video shot and 30 s of audio, each pointing back to its page or second, next to an FTS5 keyword index, because no embedding holds an invoice number exactly.
• DigUp fuses the two by turning cosine into a per-query z-score (how far a match stands out for this query), gating on it, and adding a word-coverage bonus. It beat reciprocal-rank fusion on their evals (R@1 0.88 vs 0.73) and can honestly say "nothing found". Small, author-run eval.
• It also pins llama.cpp, records the model version in the index, and checks every upgrade against stored reference vectors. Vectors from two runtimes don't mix, and nothing errors when they do.

Embedding pipelines do not feed whole documents straight into the model. They first use the embedding model's tokenizer to measure and split text into token-bounded, overlapping chunks, then embed each chunk and store the vectors. Tokenizer must match the model. Tools like marcelroed/gigatoken accelerate this preparation at corpus scale but do not create embeddings themselves.

Many asymmetric embedding models need task prefixes on inputs, and skipping them quietly degrades relevance. Each model has its own scheme: nomic (search_query:/search_document:), E5 (query:/passage:), BGE (a query instruction sentence, bare docs) — not interchangeable. OpenAI text-embedding-3-* and all-MiniLM-L6-v2 need none. Whether you add the prefix depends on the serving layer, not the model: raw endpoints (llama.cpp /v1/embeddings, HF TEI, Ollama) send bare text so it's on you, while sentence-transformers (prompt_name="query") and vendor SDKs inject it for you. Adding search_query:/search_document: to a nomic-v1.5 call lifted cosine similarity on a real query/doc pair from 0.54 to 0.60 at zero cost.

pageindex generates a semantic tree-like json index of a lengthy document to allow for reasoning based RAG without the need for vectordb.

For generating embedding locally, nomic-embed-text is a large context length text encoder that surpasses OpenAI text-embedding-ada-002 and text-embedding-3-small performance on short and long context tasks. It has a balance of speed, 8k context, and accuracy for English-centric apps. BGE-M3, Qwen3-Embedding and E5-Small are other alternatives.

alibaba/zvec by alibaba is an embedded vector database supports both spare and dense vectors, along with structured data. It can be considered the sqlite of vector databases.

yichuan-w/LEANN is a RAG focused framework focused on efficient storage with built-in chunking strategies embedding model management and MCP server. gemini

Vortex a newer format is supported by duckdb promises to be faster for random access and has zero-copy metadata, with better compression. While the support across the board is limited but worth considering for LLM/RAG based uses over parquet.

Puzer published a github recommendor that uses semantic embedding from user's github stars all client side, I found some great recommendations which I plan to use:

• pamburus/hl: A fast and powerful log viewer and processor that converts JSON logs or logfmt logs into a clear human-readable format. (⭐2657)
• samwho/spacer: CLI tool to insert spacers when command output stops (⭐1663)
• darrenburns/posting: The modern API client that lives in your terminal. (⭐11134)
• plutov/oq: Terminal OpenAPI Spec viewer (⭐943)
• wey-gu/py-pglite: PGlite wrapper in Python for testing. Test your app with Postgres just as lite as SQLite. (⭐577)

VectorChord is a faster pgvector alternative.