gavinlaw/rasst-infinisst-baseline

RASST release artifact: the InfiniSST no-RAG baseline Speech-LLM (Qwen3-Omni-30B-A3B-Instruct, LoRA r16, sampling variant, merged to a Hugging Face model directory). It is the model_infinisst_baseline asset of the RASST main_result_baseline_no_rag manifest and receives no retrieved terms; args.json holds the training arguments.

Training speech comes from GigaSpeech, which may be used only for non-commercial research and education, so these weights are released under CC BY-NC 4.0.

See the RASST repository for manifests and launch wrappers: https://github.com/luojiaxuan/RASST

Downloads last month
27
Safetensors
Model size
35B params
Tensor type
BF16
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for gavinlaw/rasst-infinisst-baseline

Finetuned
(42)
this model