4arXiv cs.CL (Computation and Language)·
Real-time sentence-level sign language translation system using SHuBERT-ByT5 with streaming hardware architecture
Researchers present a sentence-level sign language translation (SLT) system fine-tuned on How2Sign using QLoRA on a SHuBERT-ByT5 stack, achieving BLEU 15.9 and BLEURT 44.7 on the test split. The primary contribution is a hardware-aware streaming architecture that distributes camera capture to a Raspberry Pi 4B client while offloading perception and translation to a CPU/GPU backend. Latency optimizations including chunked ingestion, parallelized perception, and a sentence-boundary state machine reduce mean post-finalization response latency by 27.7% to 1.354 seconds. The system is designed for real-time deployment across diverse client devices rather than proposing a novel translation architecture.
Related guides (2)
Related events (8)
SigLIP 2: A better multilingual vision language encoder
Google releases SigLIP 2, an improved multilingual vision-language encoder model published via Hugging Face blog. The update targets better multilingual understanding and vision-language alignment compared to the original SigLIP. The post appears to cover architectural improvements and benchmark results for this encoder model, which is commonly used as a backbone in multimodal systems.
AlignAtt4LLM adapts simultaneous speech translation policy to decoder-only LLMs for IWSLT 2026
Researchers present AlignAtt4LLM, a simultaneous speech translation system for IWSLT 2026 covering English to German, Italian, and Chinese. The system cascades Qwen3-ASR for incremental transcription with Gemma-4 E4B-it for translation, applying a novel AlignAtt policy adapted for decoder-only LLMs that lack encoder-decoder cross-attention. Key contributions include explicit source span prompting, offline alignment head selection, and query/key capture to recover a usable attention-based read/write policy. The system outperforms IWSLT 2026 baselines for European language pairs in both low- and high-latency regimes.
Zero-shot image-to-text generation with BLIP-2
Hugging Face published a blog post introducing BLIP-2, a multimodal model that enables zero-shot image-to-text generation by bridging frozen image encoders and large language models via a lightweight Querying Transformer (Q-Former). The post covers the model's architecture, capabilities, and how to use it via the Hugging Face Transformers library. BLIP-2 achieves strong performance on visual question answering and image captioning tasks without task-specific fine-tuning.
HULAT2 multi-agent LangGraph system for Spanish Easy-to-Read text simplification at MER-TRANS 2026
Researchers from HULAT2-UC3M describe their submission to the MER-TRANS 2026 shared task on multilingual Easy-to-Read translation, using a LangGraph-based multi-agent workflow combining Gemini 2.5 Flash and RigoChat-7B-v2. The best run (RUN1) achieved a SARI score of 44.05 using Event-Condition-Action routing and internal quality signals, outperforming a LoRA-adapted generate-evaluate-regenerate baseline. Results show signal-guided multi-agent routing outperforms linear regeneration, while adding lexical support did not automatically improve reference-based scores.
Making LLMs even more accessible with bitsandbytes, 4-bit quantization and QLoRA
Hugging Face published a blog post detailing the integration of 4-bit quantization via bitsandbytes into the Transformers library, enabling large language models to run on consumer-grade hardware. The post covers NF4 (NormalFloat4) data type and double quantization techniques from the QLoRA paper, which together reduce memory footprint significantly while preserving model quality. It demonstrates how users can load models like LLaMA in 4-bit precision and fine-tune them using QLoRA with minimal code changes.
Accelerating Vision-Language Models: BridgeTower on Habana Gaudi2
This Hugging Face blog post covers the deployment and acceleration of BridgeTower, a vision-language model, on Intel's Habana Gaudi2 AI accelerator hardware. The piece likely benchmarks inference throughput and training performance on Gaudi2 compared to other hardware. It represents a practical infrastructure and deployment case study for multimodal models on alternative AI accelerators.
CUNI submits 1B-parameter simultaneous speech translation system to IWSLT 2026
Researchers from CUNI submit a simultaneous speech translation system to the IWSLT 2026 shared task, built on the offline Canary model with the AlignAtt policy. The system covers Czech-English and English-German/Italian translation pairs, supports 25 source and 25 target languages, and outperforms similarly sized baselines in both low- and high-latency regimes. At 1B parameters, it is positioned as a compact, multilingual, computationally efficient solution.
KIT submission to IWSLT 2026 cross-lingual voice cloning track with language tag prompting and RL fine-tuning
Researchers from KIT describe their system for the IWSLT 2026 Cross-Lingual Voice Cloning shared task, which aims to synthesize speech in a target language while preserving source-speaker identity. The system builds on FishAudio-S2-Pro, a multilingual TTS model, and introduces language tag prompting to reduce accent leakage, RL fine-tuning for intelligibility, and a reference-conditioned lexical matching method for domain-specific pronunciation. Language prompting yields the largest gains; lexical matching provides consistent improvements on matched subsets.