AI-ML
Ollama v0.20.8
RESUMEN
What's Changed ROCm: Update to ROCm 7.2.1 on Linux by @saman-amd in gemma4: fix nothink case renderer by @drifkin in * gemma4: fix compiler error on metal by @dhiltgen in
Descripción Detallada
What's Changed ROCm: Update to ROCm 7.2.1 on Linux by @saman-amd in gemma4: fix nothink case renderer by @drifkin in gemma4: fix compiler error on metal by @dhiltgen in gemma4: add nothink renderer tests by @drifkin in mlx: mixed-precision quant and capability detection improvements by @dhiltgen in mlx: add op wrappers for Conv2d, Pad, activations, trig, and masked SDPA by @dhiltgen in Revert "gemma4: add nothink renderer tests" by @drifkin in cgo: suppress deprecated warning to quiet down go build by @dhiltgen in mac: prevent generate on cross-compiles by @dhiltgen in Revert "gemma4: fix nothink case renderer" by @drifkin in launch/opencode: use inline config by @hoyyeva in gemma4: restore e2b-style nothink prompt by @drifkin in * Gemma4 on MLX by @dhiltgen in Full Changelog:
Actualización a ROCm 7.2.1 y correcciones en gemma4, incluye cambios que rompen código.
- Actualización a ROCm 7.2.1 en Linux.
- Corrección de errores en gemma4, incluyendo un error de compilación en metal.
- Se agregaron y luego se eliminaron pruebas para el renderer 'nothink'.
A quién le importa
Solo si usas gemma4 o ROCm en Linux.
Generado por IA · puede contener errores
Releases Relacionados
AI-ML
Ollama v0.32.5
## What's Changed * Fixed an MLX Metal bug that could reduce output quality for NVFP4 models, particularly Laguna. **Full Changelog**: https://github.com/ollama/ollama/compare/v0.32.4...v0.32.5
AI-ML
Ollama v0.32.4
## What's Changed - Support Laguna on Apple GPUs via the MLX engine - Quantize draft-model output heads at the requested type when creating speculative-decoding drafts. - Fixed Qwen3 MoE decoding for differently-quantized experts, plus faster packed gate/up projection (~4–9% on M5 Max). **Full
AI-ML