Downloads · 30 days
58
23% of all-time downloads
asset-7f2c9d
pnsw123/asset-7f2c9d
asset-7f2c9d is a automatic speech recognition model from pnsw123. Use it when you need speech turned into text. The card lists the license as other.
GGUF weights converted from a dialectal-Arabic fine-tune of an NVIDIA FastConformer-RNNT-with-prompt checkpoint, for use with transcribe.cpp (ggml speech-to-text inference).
All-time downloads
251
Public
Repo size
6.3 GB
Likes
1
Public
Repo makeup
Click a slice to open those files.
.gguf6.3 GB · 100%
At a glance
- Task
- Automatic Speech Recognition
- License
- other
- Access
- Public
- Created
- Sep 6, 2026
- Updated
- Sep 6, 2026
- SHA
- 8b701916
Model card
From the Hugging Face model README
GGUF conversion
GGUF weights converted from a dialectal-Arabic fine-tune of an NVIDIA FastConformer-RNNT-with-prompt checkpoint, for use with transcribe.cpp (ggml speech-to-text inference).
Source
- Base architecture / checkpoint:
nvidia/nemotron-3.5-asr-streaming-0.6b - Fine-tune:
oddadmix/nemotron-3.5-asr-arabic-dialectal(direct fine-tune, no vocab/tokenizer change — vocab size 13087, matches base)
Files
| file | dtype | size |
|---|---|---|
asset-7f2c9d-F32.gguf | float32 (reference) | ~2.4 GB |
asset-7f2c9d-F16.gguf | float16 | ~1.2 GB |
asset-7f2c9d-Q8_0.gguf | 8-bit | ~716 MB |
asset-7f2c9d-Q6_K.gguf | 6-bit | ~592 MB |
asset-7f2c9d-Q5_K_M.gguf | 5-bit | ~533 MB |
asset-7f2c9d-Q4_K_M.gguf | 4-bit | ~473 MB |
Converted with scripts/convert-parakeet.py from transcribe.cpp (the same
converter used for the upstream nemotron-3.5-asr-streaming-0.6b GGUFs),
quantized with transcribe-quantize.
License
Upstream fine-tune (oddadmix/nemotron-3.5-asr-arabic-dialectal) lists its
license as "other" with no linked terms. Base checkpoint license:
OpenMDW-1.1. Verify terms with the
fine-tune author before relying on this for anything beyond local testing.
- Task
- Automatic Speech Recognition
- License
- other
- Created
- Sep 6, 2026
- Updated
- Sep 6, 2026