Update README.md
Browse files
README.md
CHANGED
@@ -12,7 +12,7 @@ base_model:
|
|
12 |
- **Version**: MLPerf v5.1
|
13 |
- **Commit**: Close Division Commit
|
14 |
- **Supported Hardware Microarchitecture**: AMD MI300/MI325
|
15 |
-
- **Transformers**: 4.
|
16 |
- **Quark:** [0.9](https://quark.docs.amd.com/latest/install.html)
|
17 |
|
18 |
## Calibration Dataset
|
@@ -38,7 +38,7 @@ The following layers are ignored during quantization:
|
|
38 |
cd examples/torch/language_modeling/llm_ptq/
|
39 |
MODEL_DIR="mistralai/Mixtral-8x7B-Instruct-v0.1"
|
40 |
DATASET="./mlperf_data/mixtral_8x7b%2F2024.06.06_mixtral_15k_calibration_v4.pkl"
|
41 |
-
OUTPUT_DIR="
|
42 |
|
43 |
python3 quantize_quark.py --model_dir "${MODEL}" \
|
44 |
--output_dir "${OUTPUT_DIR}" \
|
|
|
12 |
- **Version**: MLPerf v5.1
|
13 |
- **Commit**: Close Division Commit
|
14 |
- **Supported Hardware Microarchitecture**: AMD MI300/MI325
|
15 |
+
- **Transformers**: 4.46.3
|
16 |
- **Quark:** [0.9](https://quark.docs.amd.com/latest/install.html)
|
17 |
|
18 |
## Calibration Dataset
|
|
|
38 |
cd examples/torch/language_modeling/llm_ptq/
|
39 |
MODEL_DIR="mistralai/Mixtral-8x7B-Instruct-v0.1"
|
40 |
DATASET="./mlperf_data/mixtral_8x7b%2F2024.06.06_mixtral_15k_calibration_v4.pkl"
|
41 |
+
OUTPUT_DIR="amd/Mixtral-8x7B-Instruct-v0.1_FP8_MLPerf_V3"
|
42 |
|
43 |
python3 quantize_quark.py --model_dir "${MODEL}" \
|
44 |
--output_dir "${OUTPUT_DIR}" \
|