nvidia/GLM-5.2-NVFP4 · Discussions

Resources

View closed (1)

Even though the accuracy seems comparable, there is significant token inflation on a per task basis

#8 opened 4 days ago by

tool_choice: "required" causes xgrammar FSM crash / infinite hang with GLM-5.2 on vLLM 0.24.0

#7 opened 6 days ago by

How is the actual programming performance?

#6 opened 24 days ago by

Some accuracy benchmark results are not as good as GLM-5.2-FP8

#5 opened about 1 month ago by

Does it support MTP?

#4 opened about 1 month ago by

Which version of VLLM should I use to run this checkpoint?

#2 opened about 1 month ago by

gameofdimension

Can we use this model with nvfp4 kv cache?

#1 opened about 1 month ago by