Cloned from monroewilliams/Qwen3.8-Flash-Next-oQ4e-fp16-mtp This model was converted from Jundot/Qwen3.8-Flash-Next-oQ4e-mtp using this script. It was not requantized, just processed to convert all bf16 parts to fp16, for better performance on M1/M2 machines. An FP16 conversion of Vontra/Qwen3.8-Flash-Next-MLX-oQ3-MTP, reprocessed with omlx-fp16-clone. Weights removed after benchmarking showed no performance benefit on M1 Ultra. FP16 was not faster than BF16 on M1 Ultra. The results show a clear split: FP16 prefill is significantly faster, but FP16 decode is slower and memory usage is higher. The omlx-fp16-clone script must promote all vision/audio passthrough tensors from BF16 to FP32 (an…
Independent publisher
Robot Haus
Robot-Haus
Models in Library4
Datasets in Library0
Models on Hugging Face10
Followers1
Models
A sensitivity-guided, mixed-precision MLX conversion of Qwen/Qwen3.8-Flash-Next, rebuilt directly from the official BF16 checkpoint with the model's matching native MTP block preserved. oQ3 uses a 3-bit affine base and spends additional precision on sensitive modules. Layer sensitivity was measured with a validated quantized calibration proxy, while every released weight was quantized from the official BF16 checkpoint. The result is a compact model with 746 higher-precision module overrides rather than a uniform 3-bit layout. The upstream tokenizer, current chat template, vision processor, generation configuration, licence, and native MTP configuration are retained. In a compatible oMLX…
An FP16 conversion of Vontra/Qwen3.8-Flash-Next-MLX-oQ3-MTP, reprocessed with omlx-fp16-clone. Weights removed after benchmarking showed no performance benefit on M1 Ultra. FP16 was not faster than BF16 on M1 Ultra. The results show a clear split: FP16 prefill is significantly faster, but FP16 decode is slower and memory usage is higher. The omlx-fp16-clone script must promote all vision/audio passthrough tensors from BF16 to FP32 (an oQ invariant — oMLX enforces this and it cannot be skipped). This produces a checkpoint with three simultaneous dtypes: integer (quantized weights), FP16 (scales and biases), and FP32 (vision/audio tensors). The mixed-dtype layout appears to prevent oMLX's…
This model was quantized using oQ (oMLX v0.6.3rc3) mixed-precision quantization.