[Quant][Kernel] Remove GPTQ Group/Dynamic Activation Ordering (#54809)

Signed-off-by: Roderick-Wu <roderickwu2003@gmail.com>
Signed-off-by: Roderick Wu <roderickwu2003@gmail.com>
Signed-off-by: mgoin <mgoin64@gmail.com>
Co-authored-by: mgoin <mgoin64@gmail.com>
This commit is contained in:
Roderick Wu
2026-09-08 12:45:12 -07:00
committed by GitHub
co-authored by mgoin
parent 1a522b6949
commit 2c9d68f80b
99 changed files with 1182 additions and 3917 deletions
+1 -3
View File
@@ -1,8 +1,6 @@
compressed-tensors, nm-testing/Mixtral-8x7B-Instruct-v0.1-W4A16-quantized, main
compressed-tensors, nm-testing/Mixtral-8x7B-Instruct-v0.1-W4A16-channel-quantized, main
compressed-tensors, nm-testing/Mixtral-8x7B-Instruct-v0.1-W8A16-quantized, main
compressed-tensors, nm-testing/test-w4a16-mixtral-actorder-group, main
gptq, TheBloke/Mixtral-8x7B-v0.1-GPTQ, main
gptq, TheBloke/Mixtral-8x7B-v0.1-GPTQ, gptq-8bit-128g-actorder_True
awq_marlin, casperhansen/deepseek-coder-v2-instruct-awq, main
compressed-tensors, RedHatAI/Llama-4-Scout-17B-16E-Instruct-quantized.w4a16, main
compressed-tensors, RedHatAI/Llama-4-Scout-17B-16E-Instruct-quantized.w4a16, main
+3 -4
View File
@@ -1,9 +1,8 @@
gptq, robertgshaw2/zephyr-7b-beta-channelwise-gptq, main
gptq, TheBloke/Llama-2-7B-GPTQ, main
gptq, TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ, main
gptq, LnL-AI/TinyLlama-1.1B-Chat-v1.0-GPTQ-4bit, main
gptq, TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ, gptq-8bit--1g-actorder_True
gptq, TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ, gptq-8bit-32g-actorder_True
gptq, TechxGenus/gemma-1.1-2b-it-GPTQ, main
gptq, TechxGenus/gemma-2b-it-GPTQ, main
compressed-tensors, nm-testing/tinyllama-oneshot-w8w8-test-static-shape-change, main
compressed-tensors, nm-testing/tinyllama-oneshot-w8-channel-a8-tensor, main
compressed-tensors, nm-testing/tinyllama-oneshot-w8a8-dynamic-token-v2, main
@@ -16,4 +15,4 @@ compressed-tensors, neuralmagic/Phi-3-medium-128k-instruct-quantized.w4a16, main
#compressed-tensors, mgoin/DeepSeek-Coder-V2-Lite-Instruct-FP8, main
awq, casperhansen/mixtral-instruct-awq, main
awq_marlin, casperhansen/mixtral-instruct-awq, main
fp8, neuralmagic/Meta-Llama-3-8B-Instruct-FP8-KV, main
fp8, neuralmagic/Meta-Llama-3-8B-Instruct-FP8-KV, main
+1 -4
View File
@@ -34,10 +34,7 @@ def test_weight_loading(vllm_runner):
"""
# MoE models need fp16.
NEEDS_FP16 = (
QUANTIZATION == "gptq"
or MODEL_NAME == "nm-testing/test-w4a16-mixtral-actorder-group"
)
NEEDS_FP16 = QUANTIZATION == "gptq"
with vllm_runner(
model_name=MODEL_NAME,
revision=REVISION,