Upload ms-swift/examples/export/quantize/gptq.sh with huggingface_hub
Browse files
ms-swift/examples/export/quantize/gptq.sh
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# OMP_NUM_THREADS=14 please Check issue: https://github.com/AutoGPTQ/AutoGPTQ/issues/439
|
| 2 |
+
OMP_NUM_THREADS=14 \
|
| 3 |
+
CUDA_VISIBLE_DEVICES=0 \
|
| 4 |
+
swift export \
|
| 5 |
+
--model Qwen/Qwen2.5-1.5B-Instruct \
|
| 6 |
+
--dataset 'AI-ModelScope/alpaca-gpt4-data-zh#500' \
|
| 7 |
+
'AI-ModelScope/alpaca-gpt4-data-en#500' \
|
| 8 |
+
--quant_n_samples 256 \
|
| 9 |
+
--quant_batch_size 1 \
|
| 10 |
+
--max_length 2048 \
|
| 11 |
+
--quant_method gptq \
|
| 12 |
+
--quant_bits 4 \
|
| 13 |
+
--output_dir Qwen2.5-1.5B-Instruct-GPTQ-Int4
|