AI-ML
Ollama v0.18.4
RESUMEN
What's Changed ggml: force flash attention off for grok by @rick-github in mlx: fix KV cache snapshot memory leak by @jessegross in * mlxrunner: schedule periodic snapshots during prefill by @jessegross
Descripción Detallada
What's Changed ggml: force flash attention off for grok by @rick-github in mlx: fix KV cache snapshot memory leak by @jessegross in mlxrunner: schedule periodic snapshots during prefill by @jessegross in doc: update vscode doc by @hoyyeva in Full Changelog:
Actualización de Ollama v0.18.4 con correcciones y mejoras.
- Desactivada la atención flash para grok.
- Corregido un problema de fuga de memoria en el caché KV.
- Programadas instantáneas periódicas durante el prellenado.
- Actualizada la documentación de VSCode.
A quién le importa
Todos los que usan Ollama.
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