GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
Reza2kn
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
Reza2kn
Model Tree
nezamisafa/whisper-persian-v4
Input Modalities
Output Modalities
Supported Functionality
Reza2kn/persian-asr-eval-v0 (FLEURS-fa)| Variant | WER ↓ | CER ↓ | clips | per-clip latency | peak VRAM |
|---|---|---|---|---|---|
| NVFP4 (this repo) | 13.30% | 5.14% | 200 | 4061 ms | 5371 MiB |
Persian text normalization for WER/CER: NFKC, ZWNJ → space, ي→ی / ك→ک, digit folding, punctuation stripping, whitespace collapse.
python
import torchimport soundfile as sffrom transformers import AutoProcessor, AutoModel repo = "Reza2kn/nezamisafa_whisper-persian-v4-NVFP4"processor = AutoProcessor.from_pretrained(repo)# Load in bfloat16 — NVFP4 weights decompress to bf16 inside CompressedLinear.model = AutoModel.from_pretrained(repo, dtype=torch.bfloat16).to("cuda").eval()(See the original nezamisafa/whisper-persian-v4 model card for arch-specific decoding boilerplate.)
llmcompressor QuantizationModifier(targets=["Linear"], scheme="NVFP4", ignore=...) →
compressed-tensors nvfp4-pack-quantized checkpoint.
Inherits the base model's license.