juanginer/Voxtral-Mini-3B-2507-AWQ4-ARG

Update benchmark table: full WER (FLEURS+CV+MLS) + Argentine + VRAM

Update benchmark table: full WER (FLEURS+CV+MLS) + Argentine + VRAM

Remove requirements/CUDA note

Update benchmark table: full WER (FLEURS+CV+MLS) + Argentine + VRAM

Update benchmark table: full WER (FLEURS+CV+MLS) + Argentine + VRAM

Update README.md

Update benchmark table: full WER (FLEURS+CV+MLS) + Argentine + VRAM

Update benchmark table: full WER (FLEURS+CV+MLS) + Argentine + VRAM

Update benchmark table: full WER (FLEURS+CV+MLS) + Argentine + VRAM

Update README.md

Fix config.json (add audio_config runtime fields + text_config.architectures) and re-add model.safetensors alias so both HF and mistral load formats work

Rename model.safetensors -> consolidated.safetensors for vLLM --config-format mistral

Replace benchmark table with re-computation notice

Replace with vLLM-compatible variant (renamed keys + expanded config)

Update README.md

Update model card (AWQ4-ARG)

Update model card (AWQ4-ARG)

Update model card (AWQ4-ARG)

Update model card (AWQ4-ARG)

Update model card (AWQ4-ARG)

Update model card (AWQ4-ARG)

Upload AWQ4-ARG

initial commit