Tag
Aleph Alpha released Kolibri, an open-weight English-German Mixture-of-Experts model with 78.1B parameters (3.46B active) and up to 1M-token context, downloadable under Apache 2.0 for sovereign, on-premises deployment in government and regulated industries.
BeeNara is a 332MB ONNX cross-encoder model for document sorting that uses split-conformal prediction to provide calibrated 'none fits' responses for uncertain classifications.
Audio8 ASR Infinite is a native streaming speech recognition model with selectable audio clock and configurable transcription delay, supporting unlimited-length audio transcription without drifting.
MePersona 是面向内容创作者的统一社媒收件箱,其 Persona 功能会学习创作者的说话风格,为 YouTube、Instagram、TikTok、X 等平台的评论和私信起草回复,并支持英西双语,所有内容需创作者确认后才发送。
Nina's Little Universe is an interactive 3D space science world for children, built as a personal project and now used in schools for bilingual education.
jialinyyzz/humanizer 是一个 12B 参数的开源模型,能将 AI 生成的英文与中文草稿改写得像人写的一样,同时保持所有数字、单位、日期和引文不变,并支持本地离线运行(提供多种 GGUF 量化版本与桌面/命令行工具)。
Dictoterix is a free language learning service that uses the dichotic method, playing native and target languages in different ears simultaneously to improve learning efficiency.
This tweet shares a curated list of English learning resources categorized by skills like listening, speaking, and reading, aimed at the Chinese online community, with bilingual support and navigation hubs.
Introduces Spark-X2.5-1.7B, a compact text generation model designed for smooth bilingual conversations in English and Chinese, with early positive reception.
Breeze TTS 2 is an open-weight, real-time text-to-speech model that supports voice design and bilingual speech, ranking first among open-weight models on the Artificial Analysis TTS leaderboard.
Skills-4-SE is a comprehensive repository providing 180+ coding skills for software engineering tasks, including requirement analysis, design, implementation, testing, and maintenance, with a web-based skill manager.
BEAR-Bench introduces a bilingual English-and-Russian benchmark for evaluating multimodal models' reasoning on text-rich professional documents, assessing 16 models and highlighting performance gaps.
This paper investigates whether bilingual Mixture-of-Experts (MoE) language models develop linguistically structured expert routing. It finds that interpretable linguistic organization emerges within MoE routing patterns, and that curriculum training influences specialization in language balance.
This paper investigates whether context-injected fine-tuning improves small language models' ability to use retrieved law for legal question answering in Bangladesh. Using Qwen3.5 at 0.8B, 2B, and 4B scales, they find fine-tuning helps at smaller scales but not at 4B, and reduces language drift.
Telco-GAIA is a bilingual, multi-modal benchmark for evaluating tool-using agents in the telecom domain, comprising 100 human-verified tasks requiring multi-hop reasoning over heterogeneous sources, with objective scoring via exact string matching.
Argelius Labs has released a bilingual search dorks list for Chinese OSINT investigations, covering methods to query company registration, procurement, courts, regulatory information, etc., aiming to help researchers conduct legitimate investigations efficiently on the Chinese internet.
The author shares their experience swapping the TTS in their voice agent to a custom model (Banter 1) designed for bilingual Arabic-English conversations, which significantly reduced perceived lag.
Horizon is an open-source AI news radar tool that automatically captures first-hand content from multiple information sources, uses AI to deduplicate, score, summarize, and generate bilingual (Chinese/English) briefings, and supports multiple deployment methods.
PolyAlign is a distribution-aware alignment framework that aligns language models to context-specific human response distributions rather than a single global style, improving naturalness and faithfulness across bilingual settings.
Linear design engineer Emil Kowalski compiled 90+ animation terms covering 12 categories. The author used Claude to create a bilingual preview site for understanding the effects.