Skip to content

update h config and add glm4.7 mtp test #4424

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
littlegy wants to merge 7 commits into
base: main
Choose a base branch
from
Open

update h config and add glm4.7 mtp test #4424

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
c17c78b
update config h and add glm4.7 mtp test
Mar 18, 2026
c9d557f
update config
littlegy Mar 18, 2026
92fba62
fix benchmark test
littlegy Mar 18, 2026
5d7f415
fix gpt-oss args
littlegy Mar 19, 2026
11424c7
update
littlegy Mar 19, 2026
8671605
update vl test config
littlegy Mar 19, 2026
a2f1446
update bencmark test
littlegy Mar 19, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion autotest/config_ascend.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,7 +82,7 @@ pytorch_quantization:
- Qwen/Qwen3-VL-8B-Instruct
- Qwen/Qwen3-VL-32B-Instruct

longtext_model:
longtext_benchmark_model:
- Qwen/Qwen3-30B-A3B

benchmark_model:
Expand Down
270 changes: 178 additions & 92 deletions autotest/config_h.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,33 +12,34 @@ device: cuda

config:
tp:
Qwen/Qwen3-235B-A22B-FP8: 4
internlm/Intern-S1: 4
Qwen/Qwen3-235B-A22B-Thinking-2507-FP8: 4
Qwen/Qwen3-30B-A3B: 2
Qwen/Qwen3-32B: 2
openai/gpt-oss-120b: 2
openai/gpt-oss-120b-BF16: 4
openai/gpt-oss-20b-BF16: 2
deepseek/DeepSeek-V3.1: 8
openai/gpt-oss-20b: 2
unsloth/gpt-oss-20b-BF16: 2
deepseek-ai/DeepSeek-V3.1: 8
Qwen/Qwen3-30B-A3B-Base: 2
JetLM/SDAR-30B-A3B-Sci: 2
moonshotai/Kimi-K2-Instruct-0905: 16
Qwen/Qwen3-235B-A22B-Thinking-2507: 8
OpenGVLab/InternVL3_5-38B: 2
Qwen/Qwen3-VL-30B-A3B-Instruct: 2
internlm/Intern-S1-Pro-FP8: 16
zai-org/GLM-5: 16
Qwen/Qwen3.5-27B: 2
Qwen/Qwen3.5-35B-A3B: 2
Qwen/Qwen3.5-122B-A10B: 4
meta-llama/Llama-4-Scout-17B-16E-Instruct: 4
meta-llama/Meta-Llama-3.1-70B-Instruct: 4
OpenGVLab/InternVL3-38B: 2
Qwen/Qwen2.5-VL-32B-Instruct: 2
deepseek-ai/DeepSeek-V2-Lite-Chat: 2
mistralai/Mixtral-8x7B-Instruct-v0.1: 2
OpenGVLab/InternVL3_5-30B-A3B: 2
zai-org/GLM-4.7-Flash: 2
google/gemma-3-27b-it: 2

dp_ep:
moonshotai/Kimi-K2-Instruct-0905:
dp: 16
ep: 16
Qwen/Qwen3-235B-A22B-Thinking-2507:
dp: 8
ep: 8
internlm/Intern-S1-Pro-FP8:
dp: 16
ep: 16

cp_tp:
Qwen/Qwen3-235B-A22B-Thinking-2507:
Expand All @@ -48,64 +49,109 @@ config:

turbomind_chat_model:
tp:
- Qwen/Qwen3-0.6B-FP8
- Qwen/Qwen3-1.7B-FP8
- Qwen/Qwen3-4B-FP8
- Qwen/Qwen3-8B-FP8
- Qwen/Qwen3-14B-FP8
- Qwen/Qwen3-235B-A22B-Thinking-2507-FP8
- Qwen/Qwen3-235B-A22B-Thinking-2507
- meta-llama/Llama-3.2-1B-Instruct
- meta-llama/Llama-3.2-3B-Instruct
- meta-llama/Meta-Llama-3.1-8B-Instruct
- meta-llama/Meta-Llama-3.1-70B-Instruct
- meta-llama/Meta-Llama-3-8B-Instruct
- internlm/internlm3-8b-instruct
- internlm/internlm3-8b-instruct-awq
- OpenGVLab/InternVL3-8B
- OpenGVLab/InternVL3-38B
- OpenGVLab/InternVL3_5-30B-A3B
- Qwen/Qwen3-0.6B
- Qwen/Qwen3-4B
- Qwen/Qwen3-8B
- Qwen/Qwen3-32B-FP8
- Qwen/Qwen3-30B-A3B
- Qwen/Qwen3-30B-A3B-FP8
- Qwen/Qwen3-32B
- Qwen/Qwen3-32B-FP8
- Qwen/Qwen3-235B-A22B-Thinking-2507
- Qwen/Qwen3-235B-A22B-Thinking-2507-FP8
- Qwen/Qwen2.5-VL-7B-Instruct
- Qwen/Qwen2.5-VL-32B-Instruct
- Qwen/Qwen1.5-MoE-A2.7B-Chat
- mistralai/Mixtral-8x7B-Instruct-v0.1
- OpenGVLab/InternVL3_5-38B
- deepseek-ai/DeepSeek-V2-Lite-Chat
- THUDM/glm-4-9b-chat
- openai/gpt-oss-120b
- openai/gpt-oss-20b

cp_tp:
- Qwen/Qwen3-235B-A22B-Thinking-2507


pytorch_chat_model:
tp:
- Qwen/Qwen3-0.6B-FP8
- Qwen/Qwen3-1.7B-FP8
- Qwen/Qwen3-4B-FP8
- Qwen/Qwen3-8B-FP8
- Qwen/Qwen3-14B-FP8
- Qwen/Qwen3-235B-A22B-Thinking-2507
- Qwen/Qwen3-235B-A22B-Thinking-2507-FP8
- meta-llama/Llama-4-Scout-17B-16E-Instruct
- meta-llama/Llama-3.2-1B-Instruct
- meta-llama/Llama-3.2-3B-Instruct
- meta-llama/Meta-Llama-3.1-8B-Instruct
- meta-llama/Meta-Llama-3.1-70B-Instruct
- meta-llama/Meta-Llama-3-8B-Instruct
- internlm/internlm3-8b-instruct
- OpenGVLab/InternVL3-8B
- OpenGVLab/InternVL3-38B
- OpenGVLab/InternVL3_5-30B-A3B
- Qwen/Qwen3-0.6B
- Qwen/Qwen3-4B
- Qwen/Qwen3-8B
- Qwen/Qwen3-32B-FP8
- Qwen/Qwen3-30B-A3B
- Qwen/Qwen3-30B-A3B-FP8
- Qwen/Qwen3-32B
- Qwen/Qwen3-32B-FP8
- Qwen/Qwen3-235B-A22B-Thinking-2507
- Qwen/Qwen3-235B-A22B-Thinking-2507-FP8
- Qwen/Qwen3-VL-8B-Instruct
- Qwen/Qwen3-VL-30B-A3B-Instruct
- OpenGVLab/InternVL3_5-38B
- unsloth/gpt-oss-120b-BF16
- THUDM/cogvlm-chat-hf
- THUDM/cogvlm2-llama3-chinese-chat-19B
- THUDM/glm-4v-9b
- THUDM/glm-4-9b-chat
- zai-org/GLM-4.7-Flash
- microsoft/Phi-3.5-vision-instruct
- microsoft/Phi-3-vision-128k-instruct
- zai-org/GLM-5
- Qwen/Qwen3.5-27B
- Qwen/Qwen3.5-35B-A3B
- Qwen/Qwen3.5-122B-A10B
- deepseek-ai/DeepSeek-V3.1
- unsloth/gpt-oss-20b-BF16
- deepseek/DeepSeek-V3.1
- moonshotai/Kimi-K2-Instruct-0905
- internlm/Intern-S1-Pro-FP8
- JetLM/SDAR-30B-A3B-Sci
- google/gemma-3-27b-it
- OpenGVLab/InternVL3_5-38B

dp_ep:
- moonshotai/Kimi-K2-Instruct-0905
- Qwen/Qwen3-235B-A22B-Thinking-2507
- internlm/Intern-S1-Pro-FP8

turbomind_vl_model:
tp:
- OpenGVLab/InternVL3-8B
- OpenGVLab/InternVL3-38B
- OpenGVLab/InternVL3_5-30B-A3B
- Qwen/Qwen2.5-VL-7B-Instruct
- Qwen/Qwen2.5-VL-32B-Instruct
- OpenGVLab/InternVL3_5-38B


pytorch_vl_model:
tp:
- OpenGVLab/InternVL3_5-38B
- OpenGVLab/InternVL3-8B
- OpenGVLab/InternVL3_5-30B-A3B
- Qwen/Qwen3-VL-8B-Instruct
- Qwen/Qwen3-VL-30B-A3B-Instruct
- THUDM/cogvlm-chat-hf
- THUDM/cogvlm2-llama3-chinese-chat-19B
- THUDM/glm-4v-9b
- microsoft/Phi-3-vision-128k-instruct
- microsoft/Phi-3.5-vision-instruct
- Qwen/Qwen3.5-27B
- Qwen/Qwen3.5-35B-A3B
- Qwen/Qwen3.5-122B-A10B
- OpenGVLab/InternVL3_5-38B


turbomind_base_model:
tp:
- Qwen/Qwen3-4B-FP8
- openai/gpt-oss-20b
- Qwen/Qwen3-8B-Base
- Qwen/Qwen3-30B-A3B-Base

pytorch_base_model:
tp:
Expand All @@ -114,94 +160,134 @@ pytorch_base_model:

turbomind_quantization:
no_awq:
- Qwen/Qwen3-0.6B-FP8
- Qwen/Qwen3-1.7B-FP8
- Qwen/Qwen3-4B-FP8
- Qwen/Qwen3-8B-FP8
- Qwen/Qwen3-14B-FP8
- Qwen/Qwen3-235B-A22B-Thinking-2507-FP8
- meta-llama/Meta-Llama-3.1-70B-Instruct
- internlm/internlm3-8b-instruct
- Qwen/Qwen3-30B-A3B
- Qwen/Qwen3-30B-A3B-FP8
- Qwen/Qwen3-32B
- Qwen/Qwen3-235B-A22B-Thinking-2507
- Qwen/Qwen3-235B-A22B-Thinking-2507-FP8
- Qwen/Qwen3-32B-FP8
- Qwen/Qwen3-30B-A3B-FP8
- Qwen/Qwen3-30B-A3B-Base
- Qwen/Qwen1.5-MoE-A2.7B-Chat
- Qwen/Qwen2.5-VL-7B-Instruct
- Qwen/Qwen2.5-VL-32B-Instruct
- OpenGVLab/InternVL3_5-30B-A3B
- deepseek-ai/DeepSeek-V2-Lite-Chat
- openai/gpt-oss-120b
- openai/gpt-oss-20b
- Qwen/Qwen3-235B-A22B-Thinking-2507
- microsoft/Phi-3-mini-4k-instruct
- THUDM/glm-4v-9b
- THUDM/glm-4-9b-chat

gptq:
- empty
no_kvint4:
- Qwen/Qwen3-0.6B-FP8
- Qwen/Qwen3-1.7B-FP8
- Qwen/Qwen3-4B-FP8
- Qwen/Qwen3-8B-FP8
- Qwen/Qwen3-14B-FP8
- Qwen/Qwen3-235B-A22B-Thinking-2507-FP8
- meta-llama/Llama-3.2-1B-Instruct
- OpenGVLab/InternVL3-2B
- OpenGVLab/InternVL3-8B
- Qwen/Qwen3-0.6B
- Qwen/Qwen3-4B
- Qwen/Qwen3-8B
- Qwen/Qwen3-32B-FP8
- Qwen/Qwen3-30B-A3B
- Qwen/Qwen3-30B-A3B-FP8
- Qwen/Qwen3-32B
- Qwen/Qwen3-32B-FP8
- Qwen/Qwen3-235B-A22B-Thinking-2507
- Qwen/Qwen3-235B-A22B-Thinking-2507-FP8
- Qwen/Qwen2.5-VL-7B-Instruct
- Qwen/Qwen2.5-VL-32B-Instruct
- Qwen/Qwen1.5-MoE-A2.7B-Chat
- Qwen/Qwen3-8B-Base
- Qwen/Qwen3-30B-A3B-Base
- deepseek-ai/DeepSeek-V2-Lite-Chat
- openai/gpt-oss-120b
- openai/gpt-oss-20b
- Qwen/Qwen3-235B-A22B-Thinking-2507
no_kvint8:
- Qwen/Qwen3-235B-A22B-Thinking-2507
- deepseek-ai/DeepSeek-V2-Chat
- Qwen/Qwen3-235B-A22B-Thinking-2507-FP8

pytorch_quantization:
awq:
- empty
- meta-llama/Llama-3.2-3B-Instruct
- meta-llama/Meta-Llama-3-8B-Instruct
- meta-llama/Meta-Llama-3.1-8B-Instruct
- internlm/internlm3-8b-instruct
- Qwen/Qwen3-0.6B
- Qwen/Qwen3-4B
- Qwen/Qwen3-8B
w8a8:
- empty
- meta-llama/Llama-3.2-1B-Instruct
- meta-llama/Meta-Llama-3-8B-Instruct
- meta-llama/Meta-Llama-3.1-8B-Instruct
- internlm/internlm3-8b-instruct
- microsoft/Phi-3-mini-4k-instruct
no_kvint4:
- meta-llama/Llama-3.2-1B-Instruct
- OpenGVLab/InternVL3-2B
- OpenGVLab/InternVL3-8B
- Qwen/Qwen3-8B-Base
- Qwen/Qwen3-0.6B-FP8
- Qwen/Qwen3-1.7B-FP8
- Qwen/Qwen3-4B-FP8
- Qwen/Qwen3-8B-FP8
- Qwen/Qwen3-14B-FP8
- Qwen/Qwen3-235B-A22B-Thinking-2507-FP8
- Qwen/Qwen3-30B-A3B-Base
- Qwen/Qwen3-0.6B
- Qwen/Qwen3-4B
- Qwen/Qwen3-8B
- Qwen/Qwen3-32B-FP8
- Qwen/Qwen3-30B-A3B
- Qwen/Qwen3-30B-A3B-FP8
- Qwen/Qwen3-32B
- Qwen/Qwen3-32B-FP8
- moonshotai/Kimi-K2-Instruct-0905
- Qwen/Qwen3-235B-A22B-Thinking-2507
- internlm/Intern-S1-Pro-FP8
- JetLM/SDAR-30B-A3B-Sci
- deepseek/DeepSeek-V3.1
- Qwen/Qwen3-235B-A22B-Thinking-2507-FP8
- Qwen/Qwen3-VL-8B-Instruct
- Qwen/Qwen3-VL-30B-A3B-Instruct
- microsoft/Phi-3-vision-128k-instruct
- microsoft/Phi-3.5-vision-instruct
- zai-org/GLM-4.7-Flash
- zai-org/GLM-5
- Qwen/Qwen3.5-27B
- Qwen/Qwen3.5-35B-A3B
- Qwen/Qwen3.5-122B-A10B
- deepseek-ai/DeepSeek-V3.1
no_kvint8:
- Qwen/Qwen3-235B-A22B-Thinking-2507
- internlm/Intern-S1-Pro-FP8
- deepseek/DeepSeek-V3.1
- zai-org/GLM-4.7-Flash
- zai-org/GLM-5
- deepseek-ai/DeepSeek-V3.1
- Qwen/Qwen3.5-27B
- Qwen/Qwen3.5-35B-A3B
- Qwen/Qwen3.5-122B-A10B
- Qwen/Qwen3-235B-A22B-Thinking-2507-FP8

longtext_model:
longtext_benchmark_model:
- Qwen/Qwen3-30B-A3B
- Qwen/Qwen3-235B-A22B-Thinking-2507
- zai-org/GLM-5
- Qwen/Qwen3.5-27B
- Qwen/Qwen3.5-35B-A3B
- Qwen/Qwen3.5-122B-A10B

benchmark_model:
- meta-llama/Meta-Llama-3-1-8B-Instruct
- meta-llama/Meta-Llama-3-1-70B-Instruct
- Qwen/Qwen3-32B
- meta-llama/Meta-Llama-3.1-8B-Instruct
- meta-llama/Meta-Llama-3.1-70B-Instruct
- Qwen/Qwen3-30B-A3B
- Qwen/Qwen3-235B-A22B-Thinking-2507
- Qwen/Qwen2.5-72B-Instruct
- openai/gpt-oss-120b
- openai/gpt-oss-20b
- unsloth/gpt-oss-20b-BF16
- unsloth/gpt-oss-120b-BF16
- zai-org/GLM-5
- Qwen/Qwen3.5-27B
- Qwen/Qwen3.5-35B-A3B
- Qwen/Qwen3.5-122B-A10B
- google/gemma-3-27b-it

evaluate_model:
- Qwen/Qwen3-32B
- Qwen/Qwen3-32B-FP8
- Qwen/Qwen3-30B-A3B
- Qwen/Qwen3-30B-A3B-FP8
- Qwen/Qwen3-235B-A22B-Thinking-2507
- Qwen/Qwen3-235B-A22B-Thinking-2507-FP8
- openai/gpt-oss-120b
- unsloth/gpt-oss-120b-BF16
- deepseek/DeepSeek-V3.1
- moonshotai/Kimi-K2-Instruct-0905
- internlm/Intern-S1-Pro-FP8
- JetLM/SDAR-30B-A3B-Sci
- deepseek-ai/DeepSeek-V3.1
- zai-org/GLM-5
- Qwen/Qwen3.5-27B
- Qwen/Qwen3.5-35B-A3B
- Qwen/Qwen3.5-122B-A10B

mllm_evaluate_model:
- OpenGVLab/InternVL3_5-38B
Expand Down
Loading
Loading