PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度
update supported scheme readme (#1257)
Wenhua Cheng6a87a651075c
diff --git a/README.md b/README.mdindex b2309a975..69e81c458 100644--- a/README.md+++ b/README.md@@ -217,6 +217,21 @@ ar.quantize_and_save(output_dir="./qmodel", format="auto_round")</details>+### Supported Schemes+<details>+> Gray indicates the absence of a kernel or the presence of only an inefficient/reference kernel. BF16 is mainly for AutoScheme++| Format | Supported Schemes |+|:-------------------|:--------------------------------------------------------------------------------------------------------------------------------------------------------------------------|+| **auto_round** | W4A16(Recommended), W2A16, W3A16, W8A16, W2A16G64, W2A16G32, `MXFP4`, `MXFP8`, `MXFP4_RCEIL`, `MXFP8_RCEIL`, `NVFP4`, `FPW8A16`, `FP8_STATIC`, `BF16` |+| **auto_awq** | W4A16(Recommended), BF16 |+| **auto_gptq** | W4A16(Recommended), W2A16, W3A16, W8A16, W2A16G64, W2A16G32,BF16 |+| **llm_compressor** | NVFP4(Recommended), `MXFP4`, `MXFP8`, `FPW8A16`, `FP8_STATIC` |+| **gguf** | GGUF:Q4_K_M(Recommended), GGUF:Q2_K_S, GGUF:Q3_K_S, GGUF:Q3_K_M, GGUF:Q3_K_L, GGUF:Q4_K_S, GGUF:Q5_K_S, GGUF:Q5_K_M, GGUF:Q6_K, GGUF:Q4_0, GGUF:Q4_1, GGUF:Q5_0, GGUF:Q5_1,GGUF:Q8_0 |+| **fake** | `all schemes (only for research)` |+</details>++### Adaptive Schemes (Experimental Feature)AutoScheme provides an automatic algorithm to generate adaptive mixed bits/data-type quantization recipes.Please refer to the [user guide](https://github.com/intel/auto-round/blob/main/docs/step_by_step.md#autoscheme) for more details on AutoScheme.diff --git a/docs/step_by_step.md b/docs/step_by_step.mdindex 654d55d50..9c77d15d8 100644--- a/docs/step_by_step.md+++ b/docs/step_by_step.md@@ -149,16 +149,18 @@ adopted within the community, **only 4-bits quantization is supported**. Please**LLM-Compressor Format**: **NVFP4, MXFP4(kernel in WIP), MXFP8 are supported**. Please set `--format llm_compressor`#### Format and scheme support matrix-> Italics indicates the absence of a kernel or the presence of only an inefficient/reference kernel.--|export format | supported scheme |-|--------------|------------------|-|**auto_round** | W4A16, W2A16, W3A16, W8A16, *MXFP4*, *MXFP8*, NVFP4, FPW8A16, W2A16G64, W2A16G32, *FP8_STATIC*, BF16|-|**auto_awq / auto_round:auto_awq** | W4A16|-|**auto_gptq / auto_round:auto_gptq / auto_round:gptqmodel**|W4A16, W2A16, W3A16, W8A16, BF16, W2A16G64, W2A16G32|-|**llm_compressor / auto_round:llm_compressor** | *MXFP4*, *MXFP8*, NVFP4, FPW8A16, *FP8_STATIC* |-|**gguf** | GGUF:Q4_0, GGUF:Q4_1, GGUF:Q5_0, GGUF:Q5_1, GGUF:Q2_K_S, GGUF:Q3_K_S, GGUF:Q3_K_M, GGUF:Q3_K_L, GGUF:Q4_K_S, GGUF:Q4_K_M, GGUF:Q5_K_S, GGUF:Q5_K_M, GGUF:Q6_K, GGUF:Q8_0 |-|**fake** | all scheme|+> Gray indicates the absence of a kernel or the presence of only an inefficient/reference kernel. BF16 is mainly for AutoScheme+++| Format | Supported Schemes |+|:---|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------|+| **auto_round** | W4A16, W2A16, W3A16, W8A16, W2A16G64, W2A16G32, `MXFP4`, `MXFP8`, `MXFP4_RCEIL`, `MXFP8_RCEIL`, `NVFP4`, `FPW8A16`, `FP8_STATIC`, `BF16` |+| **auto_awq** | W4A16, BF16 |+| **auto_gptq** | W4A16, W2A16, W3A16, W8A16,W2A16G64, W2A16G32, BF16 |+| **llm_compressor** | NVFP4, `MXFP4`, `MXFP8`, `FPW8A16`, `FP8_STATIC` |+| **gguf** | GGUF:Q4_K_M, GGUF:Q2_K_S, GGUF:Q3_K_S, GGUF:Q3_K_M, GGUF:Q3_K_L, GGUF:Q4_K_S, GGUF:Q5_K_S, GGUF:Q5_K_M, GGUF:Q6_K, GGUF:Q4_0, GGUF:Q4_1, GGUF:Q5_0, GGUF:Q5_1,GGUF:Q8_0 |+| **fake** | `all schemes (only for research)` |+### Hardware CompatibilityCPU, Intel GPU, HPU and CUDA for both quantization and inference.