Hugging Face's logo

danish-foundation-models
/
munin-7b-alpha

Text Generation
Transformers
Safetensors
Danish
mistral
pretrained
Inference Endpoints
text-generation-inference
Model card Files Community
3

Any reason to set `use_cache` to `false`

#3
by sorenmulli - opened
Discussion
sorenmulli

With this setting, the default generation config means slooooooow decoding. Any reason I am missing? :)

Mistral has it as true.

KennethEnevoldsen
 Danish Foundation Models org

@synquid ?

Sign up or log in to comment