AI-ML
Ollama v0.20.1
RESUMEN
What's Changed bench: add prompt calibration, context size flag, and NumCtx reporting by @dhiltgen in model/parsers: fix gemma4 arg parsing when quoted strings contain " by @drifkin in * ggml: skip cubl
Descripción Detallada
What's Changed bench: add prompt calibration, context size flag, and NumCtx reporting by @dhiltgen in model/parsers: fix gemma4 arg parsing when quoted strings contain " by @drifkin in ggml: skip cublasGemmBatchedEx during graph reservation by @jessegross in gemma4: enable flash attention by @dhiltgen in ggml: fix ROCm build for cublasGemmBatchedEx reserve wrapper by @jessegross in model/parsers: rework gemma4 tool call handling by @drifkin in Full Changelog:
Actualización de Ollama v0.20.1 con mejoras en el análisis de argumentos y soporte para atención flash.
- Se añade calibración de prompts, tamaño de contexto y reporte de NumCtx en bench.
- Se corrige el análisis de argumentos en gemma4 para cadenas entre comillas.
- Se habilita la atención flash en gemma4 y se reestructura el manejo de llamadas a herramientas.
A quién le importa
Todos los que usan Ollama para modelos de IA.
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