Commit History
3 days ago
-
Card: tool calls on litert-lm 0.18.0 — model.litertlm re-packed 2026-10-14 (llm_model_type qwen2p5; the chat template now accepts tools and renders tool results; metadata only); GPU executor defaults note
Verified signaturec98c88a -
Re-pack model.litertlm: llm_model_type qwen2p5, a chat template that accepts tools and renders tool results (metadata only; tool calls parse and tool results reach the model on litert-lm 0.18.0)
Verified signaturea553da4
25 days ago
-
Chat template: accept the 0.18 content-parts form (string form unchanged); weights, tokenizer and executor metadata byte-identical
Verified signature3190c73
-
Card: base_model_relation: quantized (so the conversion lists under the base model's Quantizations, not Finetunes)
Verified signaturef5c14ec
-
Link the card back to its collection and the request box
Verified signature3b658a5
-
Add desktop LiteRT-LM CLI section (import/run/serve)
Verified signature13c1bb0
-
Card: add 'Reproduce (official tools only)' section
Verified signature172e92d -
Add the blockwise-128 int4 recipe (reproduce with stock litert-torch)
Verified signaturebb4fa29 -
Rebuild via stock litert-torch (recipe.json, no custom code); same blockwise-128 int4, GSM8K 77%
Verified signature1e553b4 -
Card: blockwise-128 + cache2048 specs (iPhone ~27 tok/s, GSM8K 77%)
Verified signature24b3b76 -
Switch to blockwise-128 int4 + cache2048 (iPhone 27 tok/s, GSM8K 77% parity)
Verified signature07c5809 -
Update card: blockwise int4 + GSM8K parity table
Verified signature9f89d30 -
Replace channelwise int4 with blockwise int4 + cache4096 (GSM8K 79% ~ bf16 75%)
Verified signature5a6d84a
-
Upload model.litertlm with huggingface_hub
Verified signature247ba94 -
Upload README.md with huggingface_hub
Verified signature792b4b3 -
initial commit
Verified signature80847f8