Yazar | SHA1 Mesaj | Tarih |
---|---|---|
|
00503b9fc1 feat: non-uniform quantization via `compressed-tensors` for llama | 5 ay önce |
|
ee2c5d34da feat: add fp8 channel-wise weight quantization support | 5 ay önce |
|
98cb1c4cd1 feat: support fp8 via `llm-compressor` | 5 ay önce |
|
e2dbe5f05c feat: add sparse marlin for compressed tensors | 6 ay önce |
|
aba03b4756 feat: dynamic per-token activation quantization | 6 ay önce |