Sage

Clases de Sage

• Hugging Face Transformers

  Transformers de Hugging Face: usar y afinar modelos sin empezar de cero

  Transformers es la librería que te deja cargar un modelo entrenado por otros y usarlo en minutos, o seguir entrenándolo con tus propios datos. La diferencia entre entrenar desde cero y afinar (fine-tuning) es la diferencia entre meses de cómputo y una tarde con tu laptop.

  3 proyectos propuestos

• Hugging Face Transformers 5.20

  Special surprise including the new model Beam from Reflection AI :partyingface:. <img width="2912" height="1440" alt="GeminiGeneratedImageox4hjlox4hjlox4h" src=" />. Beam is a token-efficient, frontier-scale model focused on coding and agentic use cases. It is built for teams that want to run a powerful model on their own hardware, infrastructure, or cloud, and customize it to their specific needs. Links: Documentation. Fix small bug in auto mappings.

• Hugging Face Transformers 5.19

  <img width="2716" height="2308" alt="image" src=" />. EmbeddingGemma 2 is a multimodal embedding model from Google built on the Gemma 4 architecture. It encodes text, images, audio, and video, individually or combined in one input, into a shared 768-dimensional vector space for cross-modal retrieval, semantic similarity, clustering, and classification. It uses Matryoshka Representation Learning, so embeddings can be truncated to 512, 256, or 128 dimensions.

• Hugging Face Transformers 5.18

  <img width="1680" height="900" alt="image" src=" />. Nemotron 3 Diarization is an open-weight streaming speaker diarization model designed to determine "who spoke when" in real-world audio. It supports both streaming and offline inference, handles up to eight speakers, and orders speaker outputs by each speaker's first arrival in the input audio. The model uses the Arrival-Order Speaker Cache (AOSC) 1 and FIFO queue introduced for Streaming Sortformer 1, 2.

• vLLM 0.30

  This release features 762 commits from 315 contributors (104 new)!. New models: DeepSeek-V4.1-Flash (#56214, #56228, #56208) with the whole KV stored in MXFP8 through the FlashMLA V4.1 record on SM100 , DeepGEMM Mega-mHC , and async Engram prefetch with Engram DP sharding ; DeepSeek-V4-Flash-Vision-Exp , also on ROCm and with LoRA ; GLM-5.3-Flash with EPLB ; K2-Horizon ; Cohere Compass ; Bailing V3 VL ; Nanbeige4.2 via the Transformers backend ;

• Hugging Face Transformers 5.17

  <img width="1503" height="827" alt="image" src=" />. Hy4-Preview is a 780B-parameter mixture-of-experts language model that activates 49B parameters per. token. Each MoE layer holds 256 routed experts plus one always-active shared expert and routes every. token to 8 of them. The context window is 1M tokens. The architecture combines four features:. Multi-head Latent Attention (MLA) compresses keys and values into a low-rank latent.

• vLLM 0.29

  This release features 594 commits from 277 contributors (91 new)!. Model Runner V2 is now the default for all models , completing the rollout that began with pooling models . MRV2 also gained CUDA graph memory profiling for KV cache auto-sizing , batch-sharded sampling that cuts per-step logits memory by 1/TP , prompt embeds , extracthiddenstates speculation , padded FULL cudagraph dispatch for uniform decode under spec decode , and DP-sync skipping before EAGLE/MTP draft prefill .