akash-coded

akash-coded/nanorag

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

0 stars 0 forks 0 watchers Python Other
bedrock bm25 evaluation fts5 hybrid-search information-retrieval jupyter-notebooks llm-as-a-judge llm-evaluation python rag reranking retrieval-augmented-generation sqlite teaching-materials vector-search
View on GitHub Website
16 Open Issues Need Help Last updated: Aug 31, 2026

Open Issues Need Help

View All on GitHub
[E4] A production ANN backend behind the same Hit interface 18 days ago
help wanted type: extension area: retrieval
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[E3] Replace LSA with a real embedding model 18 days ago
help wanted type: extension area: retrieval
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[E1] Ingest MultiHop-RAG behind the corpus adapter 18 days ago
help wanted type: extension area: toolkit
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[ext] Corrective retrieval — act on an insufficiency verdict instead of abstaining 18 days ago
help wanted type: extension area: evaluation area: agent needs: eval-numbers
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[ext] A replay eval slice from real cohort questions — for distribution, not volume 19 days ago
help wanted type: extension type: discussion-followup area: evaluation
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[ci] A measurement-completeness check for PRs that touch the retrieval path 19 days ago
good first issue type: chore type: discussion-followup area: ci
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[ext] A stopping rule with a cost ceiling, and the quality/spend curve behind it 19 days ago
help wanted type: extension area: agent area: cost needs: eval-numbers
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[ext] Measure position sensitivity against a real model, not the extractive reader 19 days ago
help wanted type: extension needs: eval-numbers area: context
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[ext] Semantic answer cache with a measured false-hit budget 19 days ago
help wanted type: extension area: cost needs: eval-numbers
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[ext] Total cost of ownership model — the 69% of the bill that is not generation tokens 19 days ago
help wanted type: extension area: cost needs: eval-numbers
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[chore] Add a real ANN backend behind the same Hit interface 19 days ago
help wanted type: chore area: retrieval
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[ext] Adversarial and robustness eval set — including prompt injection through retrieved content 19 days ago
help wanted type: extension area: evaluation
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[docs] The README's three contradicting findings need a 'how to re-test this' section 19 days ago
good first issue type: docs area: docs
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[ext] Route the fusion weight α by query class 19 days ago
help wanted type: extension area: retrieval needs: eval-numbers
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[ext] Solve abstention — no retrieval-score threshold works 19 days ago
help wanted type: extension area: evaluation needs: eval-numbers
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search
[ext] Add HyDE — hypothetical document embeddings 19 days ago
help wanted good first issue type: extension area: retrieval needs: eval-numbers
akash-coded/nanorag
0

Runnable retrieval / RAG / evaluation curriculum — 10 notebooks and a toolkit that run entirely in memory, with an eval gate in CI

Python
#bedrock#bm25#evaluation#fts5#hybrid-search#information-retrieval#jupyter-notebooks#llm-as-a-judge#llm-evaluation#python#rag#reranking#retrieval-augmented-generation#sqlite#teaching-materials#vector-search