David Corvoysier

dacorvo

60 12

https://www.kaizou.org

dacorvo

AI & ML interests

Quantization

Recent Activity

commentedon an article about 1 hour ago

funes: Local Memory for Coding Agents, Built on Lance

updated a dataset about 16 hours ago

dacorvo/transformers-gh-memory

updated a Space 2 days ago

dacorvo/transformers-gh-index

View all activity

Organizations

commented on funes: Local Memory for Coding Agents, Built on Lance about 1 hour ago

Yes, partially. If the agent supports STDIN MCP servers, the read functions are readily available. For the automatic indexing, you will need to write an adapter following a contract that will be available in the next release (1.4).

updated a dataset about 16 hours ago

dacorvo/transformers-gh-memory

Updated 3 minutes ago • 6.25k

updated a Space 2 days ago

transformers-gh-index

🧠

Keeps a funes memory of a GitHub repo's issues and PRs fresh

updated a bucket 3 days ago

huggingface/funes

12.5 GB

published a dataset 4 days ago

dacorvo/transformers-gh-memory

Updated 3 minutes ago • 6.25k

published a Space 4 days ago

transformers-gh-index

🧠

Keeps a funes memory of a GitHub repo's issues and PRs fresh

commented on Give Your Coding Agents a Memory You Own 6 days ago

longmemeval does not really need semantic search: BM25 already gives the same ratio you mentioned IIRC, so for this use case this is pretty much expected that it is not worth paying the cost of evaluating the embeddings (this extra cost you see at indexing and recall time). funes shines when the information is actually difficult to find.