SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

[AutoGPTQ] Add correct installation of GPTQ library + fix slow tests (#25713)

Browse Source 
* add correct installation of GPTQ library

* update tests values
This commit is contained in:
Younes Belkada
2023-08-24 14:57:16 +02:00
committed by GitHub
parent 2febd50614
commit 584eeb5387
2 changed files with 4 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
tests/quantization/gptq/test_gptq.py
View File

@@ -87,7 +87,8 @@ class GPTQTest(unittest.TestCase):
EXPECTED_OUTPUTS = set()
EXPECTED_OUTPUTS.add("Hello my name is John and I am a professional photographer. I")
EXPECTED_OUTPUTS.add("Hello my name is John and I am a very good looking man.")
EXPECTED_OUTPUTS.add("Hello my name is Alyson and I am a professional photographer")
EXPECTED_OUTPUTS.add("Hello my name is Alyson, I am a student in the")
EXPECTED_OUTPUTS.add("Hello my name is Alyson and I am a very sweet,")
# this seems a little small considering that we are doing 4bit quant but we have a small model and ww don't quantize the embeddings
EXPECTED_RELATIVE_DIFFERENCE = 1.664253062
@@ -215,7 +216,7 @@ class GPTQTest(unittest.TestCase):
self.assertEqual(self.quantized_model.config.quantization_config.disable_exllama, True)
# we need to put it directly to the gpu. Otherwise, we won't be able to initialize the exllama kernel
quantized_model_from_saved = AutoModelForCausalLM.from_pretrained(
tmpdirname, quantization_config=GPTQConfig(disable_exllama=False, bits=6), device_map={"": 0}
tmpdirname, quantization_config=GPTQConfig(disable_exllama=False, bits=4), device_map={"": 0}
)
self.assertEqual(quantized_model_from_saved.config.quantization_config.disable_exllama, False)
self.assertEqual(quantized_model_from_saved.config.quantization_config.bits, self.bits)