biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[CI] Prune tests/models/decoder_only/language/* tests (#9940)

Browse Source 
Signed-off-by: mgoin <michael@neuralmagic.com>
This commit is contained in:
Michael Goin
2024-11-05 16:02:23 -05:00
committed by GitHub
parent b9c64c0ca7
commit 02462465ea
9 changed files with 70 additions and 270 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
tests/models/decoder_only/language/test_gptq_marlin.py
View File

@@ -22,24 +22,11 @@ os.environ["TOKENIZERS_PARALLELISM"] = "true"
MAX_MODEL_LEN = 1024
MODELS = [
# act_order==False, group_size=channelwise
("robertgshaw2/zephyr-7b-beta-channelwise-gptq", "main"),
# act_order==False, group_size=128
("TheBloke/Llama-2-7B-GPTQ", "main"),
# act_order==True, group_size=128
("TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ", "main"),
# act_order==True, group_size=64
("TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ", "gptq-4bit-64g-actorder_True"),
# act_order==True, group_size=32
("TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ", "gptq-4bit-32g-actorder_True"),
# 8-bit, act_order==True, group_size=channelwise
("TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ", "gptq-8bit--1g-actorder_True"),
# 8-bit, act_order==True, group_size=128
("TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ", "gptq-8bit-128g-actorder_True"),
# 8-bit, act_order==True, group_size=32
("TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ", "gptq-8bit-32g-actorder_True"),
# 4-bit, act_order==True, group_size=128
("TechxGenus/gemma-1.1-2b-it-GPTQ", "main")