3 projects
thinletter-eq
Renamed to thinembed. This release only installs thinembed; the import name eq and the command eq are unchanged.
thinembed
ThinEmbed (formerly thinletter-eq). Measure a compressed query encoder against your own document index: import, synthetic queries, fp32 cache, llama.cpp compile, paired evaluation.
thinctx
ThinCtx: inference runtime and OpenAI-compatible server for the ThinCtx models (a small GPU memory for the context that grows 20 to 30 times more slowly than the original model's KV cache).
