Mixtral-8x7B-Instruct-v0.1-WMXFP4FP8-AMXFP4FP8-AMP-KVFP8

  • Introduction

    This model was created by applying Quark with calibration samples from Pile dataset.

  • Quantization Stragegy

    • Quantized Layers: All linear layers excluding "lm_head", "*.gate"
    • Weight: Auto Mixed Precision quantized by Quark, each weight has either quantization scheme in candidates of
      • FP8 symmetric per-tensor
      • OCP Microscaling (MX) FP4
    • Activation: Auto Mixed Precision quantized by Quark, each activation input has the same quantization scheme with weight, i.e., in candidates of
      • FP8 symmetric per-tensor
      • OCP Microscaling (MX) FP4
    • KV Cache: FP8 symmetric per-tensor
  • Quick Start

  1. Download and install Quark
  2. [TODO] We will provide example script(s) to run auto mixed precision (AMP) quantizations later.

Deployment

The Quark quantized Auto Mixed Precision (AMP) models are now supported to be easily deployed in vLLM backend (vLLM-compatible).

Evaluation

The quantization evaluation results are conducted in pseudo-quantization mode, which may slightly differ from the actual quantized inference accuracy. These results are provided for reference only.

Evaluation scores

Quant scheme arc challenge (โ†‘)
(acc)
gsm8k (โ†‘)
(strict-match)
mmlu (โ†‘)
(acc)
wikitext (โ†“)
(word_perplexity)
winogrande (โ†‘)
(acc)
absolute value recovery rate absolute value recovery rate absolute value recovery rate absolute value recovery rate absolute value recovery rate
FP16 0.6271 100.0% 0.6384 100.0% 0.6893 100.0% 5.7259 100.0% 0.7664 100.0%
FP8 0.6220 99.2% 0.6384 100.0% 0.6816 98.9% 5.8092 98.6% 0.7830 102.2%
AMP 0.6237 99.5% 0.6133 96.1% 0.6765 98.1% 6.2020 92.3% 0.7569 98.8%
MXFP4 0.5922 94.4% 0.4845 75.9% 0.6364 92.3% 7.0935 80.7% 0.7474 97.5%

License

Modifications copyright(c) 2024 Advanced Micro Devices,Inc. All rights reserved.

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.

Downloads last month
13
Safetensors
Model size
37B params
Tensor type
BF16
ยท
F8_E4M3
ยท
U8
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for amd/Mixtral-8x7B-Instruct-v0.1-WMXFP4FP8-AMXFP4FP8-AMP-KVFP8

Quantized
(41)
this model

Collection including amd/Mixtral-8x7B-Instruct-v0.1-WMXFP4FP8-AMXFP4FP8-AMP-KVFP8