New
Everything new in AI as it appears: launches announced by the labs, brand-new models, datasets, papers and open-source tools. Checked every minute or two. Launches feed · Trending · News
Live now
- Connecting…
Just launched
Yesterday
- Model
- Dataset
- Modelseiginomikata587/GGvlist style woman
Nvlist Style char
- Model
- Model
- Model
- Model
- Model
- Model
- Model
- Model
- Model
- Datasetsidchalamala/aether-benchmark
Aether benchmark 110 tasks: an English SOP, the owner's policy, sheet declarations, and a reference Aether program. They cover 25 business types and 52 categories. 30 tas
- Model
- Model
- Model
- DatasetTUaxhuiax/huggingface_filesystem_fetch_excel_2965_src_x7q3m2
Support Conversation Logs (raw export) Synthetic multi-turn customer-support conversations exported from the support platform. Each line of raw_conversations.jsonl is one
- Dataset
- Model
- Dataset
- Model
- Dataset
- Dataset
- Dataset
- Datasetsidchalamala/aether-sop-programs
Aether synthetic SOP-to-program dataset Pairs of plain-English business procedures (SOPs) and programs in Aether, a small language for checkable business procedures. Each
- Dataset
- Dataset
- Toolxgen-dex-cli
XGEN DexOn CLI — 터미널에서 XGEN 에이전트와 대화하고, 이 컴퓨터의 셸·파일·MCP 를 에이전트에게 빌려 줍니다.
- DatasetTUaxhuiax/huggingface_filesystem_fetch_2957-gsm8k-llama-factory-sharegpt-a7c3d9
GSM8K — LLaMA-Factory sharegpt Small GSM8K sample (64 examples) converted to the ShareGPT conversation format used by LLaMA-Factory. Rows: 64 Format: sharegpt
- DatasetTUaxhuiax/huggingface_filesystem_fetch_2957-gsm8k-openrlhf-filtered-a7c3d9
GSM8K — OpenRLHF filtered subset Filtered GSM8K subset (180 examples) prepared for an OpenRLHF SFT run. Alpaca-style instruction/input/output fields. Rows: 180 Format: al
- Model
- DatasetTUaxhuiax/huggingface_filesystem_fetch_2957-gsm8k-trl-ready-a7c3d9
GSM8K — TRL SFT ready Community conversion of GSM8K (128 examples) already formatted for TRL's SFTTrainer chat template. Download count is low because this is a personal
- Model
- Datasetplaywithmino/SAVE
SAVE: paper split This repository contains the paper's selected 35,768 synthetic scenes: 32,768 training, 1,000 validation and 2,000 test scenes. It is a subset of the la
- Dataset
- Model
- Model
- ModelYunmiyun_UwU/MOSS architecture converter to int8_convrot
A script that allows you to convert a bf16 checkpoint of the MOSS architecture into an int8 ConvRot format with a single click. It even supports non-standard attention bl
- Dataset
- Model
Headlines and links come from the labs' own blogs, Hacker News and the GDELT news index, and from the sources our index already follows. We show a short excerpt and always link to the original. A launch reported by a news site rather than the lab is marked “Reported” until the lab itself announces it.