Daniël de Kok 4594e6faba Add support for Marlin-quantized models ... This change adds support for Marlin-quantized models. Marlin is an FP16xINT4 matmul kernel, which provides good speedups decoding batches of 16-32 tokens. It supports quantized models with symmetric quantization, groupsize -1 or 128, and 4-bit. Tested with: - Llama 2 - Llama 3 - Phi 3		2024-06-06 13:16:52 +02:00
..
consuming_tgi.md	chore: add pre-commit (#1569)	2024-02-16 11:58:58 +01:00
gated_model_access.md	Update documentation version to 2.0.4 (#1980)	2024-05-31 16:03:24 +02:00
launcher.md	Add support for Marlin-quantized models	2024-06-06 13:16:52 +02:00
monitoring.md	docs: Fix grafana dashboard url (#1925)	2024-05-21 13:12:14 -04:00
non_core_models.md	chore: add pre-commit (#1569)	2024-02-16 11:58:58 +01:00
preparing_model.md	chore: add pre-commit (#1569)	2024-02-16 11:58:58 +01:00
safety.md	Pickle conversion now requires `--trust-remote-code`. (#1704)	2024-04-05 13:32:53 +02:00
train_medusa.md	fix small typo and broken link (#1958)	2024-05-27 11:31:06 -04:00
using_cli.md	chore: add pre-commit (#1569)	2024-02-16 11:58:58 +01:00
using_guidance.md	feat: prefer huggingface_hub in docs and show image api (#1844)	2024-05-02 16:56:24 +02:00
visual_language_models.md	feat: prefer huggingface_hub in docs and show image api (#1844)	2024-05-02 16:56:24 +02:00