Tes super menarik oleh atom_chat_hq

Menunjukkan bahwa kuantisasi yang lebih presisi tidak serta merta memberikan hasil yang lebih baik. Di sini, Qwen 3.8 27B Q8 (model versi terkuantisasi 8-bit) terkadang cocok atau mengalahkan Q4 (terkuantisasi 4-bit) pada tugas voxel yang sama. Mereka akhirnya

Rohan PaulAug 2415

NOTICIAS IA🚨: Han liberado una IA sin censura que funciona en tu ordenador

Accede al vídeo aquí:  🔥 Los modelos open source tienen cosas muy buenas, pero también riesgos importantes, y esta semana hemos visto las dos caras. Qwen 3.8 sorprende con solo

Inteligencia ArtificialAug 2412

Atomic Chat HQ rented a B200 cluster

Fired up Alibaba's 2.4-trillion parameter Qwen 3.8-Max model, and told it to build a full Call of Duty clone from a single prompt. The media thought that was the story. It was not. The crew thought that was the story. It was not. The story

Pascual ⚡Aug 24

Four frontier models took mazebench today and all four scored 0%.

The one that beat them scored 1% The account posting this built the benchmark, so it is the source and not a screenshot of a screenshot. Ox Alpha zero. Grok 4.6 zero. GLM 5.3 zero. Qwen 3.8 Max zero. Gemini 3.7

StarHazeAug 243