SuperAPI AI Directory
Partner with us中文AI newsFavorites

All AI tools

35,738 results · Page 282 of 993

Live Portrait

AI images

Live Portrait is a video-driven portrait animation framework that can transform a single static portrait photo into a realistic dynamic video. It was developed by researchers from Kuaishou Technology, the University of Science and Technology of China, and Fudan University, and is based on an implicit keypoint framework and a vid2vid model. It can extract facial expressions and head pose movements from a driving video and transfer them to a source portrait. The tool supports multi-style animation, cross-reenactment, self-reenactment, precise control of eye opening and closing and lip position, as well as video-to-video face editing. It uses stitching and retargeting modules to achieve fine-grained control of expressions and motion through a small MLP. Live Portrait achieves a generation speed of 12.8ms on an RTX 4090 GPU using PyTorch, and its training used approximately 69 million high-quality frames. Users can upload a portrait image and a driving video to generate an animation in three steps. The tool offers a free trial and is suitable for portrait animation, video editing, and many other scenarios, and can also be fine-tuned for animal animation.

Read more

livemy.app

AI coding

livemy.app is a deployment platform for non-ops users that can deploy web apps, Docker containers, and open-source tools generated from GitHub repositories, ZIP archives, or AI coding tools (such as Cursor and Lovable) to production with one click, with no server configuration required. It supports MCP integration, allowing AI agents to complete deployments directly and return a URL. It offers a free plan (projects stay online for 24 hours after each deployment) and paid plans (such as $10/month Maker including 5 projects, unlimited traffic, and 24/7 online). The platform automatically detects the tech stack, configures SSL, monitors, and automatically restarts, making it suitable for founders, designers, and indie developers to launch projects quickly.

Read more

Llamaberry-Chain-of-Thought-Reasoning-in-AI

AI developer platforms

Llamaberry is a multi-turn Chain of Thought (CoT) reasoning system implemented based on the Llama 3.1 70B model on the Groq platform. It puts a user's question through three rounds of reasoning in sequence: the first round performs preliminary brainstorming, the second round refines and improves upon the first round, and the third round further deepens it, finally synthesizing the results of the three rounds into a conclusion. The project provides a user-friendly interface that can display the thinking process of each round and simultaneously gives answers in both concise and detailed styles; in addition, it also displays metrics such as reasoning time. Using this project requires a Groq API key.

Read more

LLM-Dojo

AI coding

LLM-Dojo is a lightweight post-training framework for large language models, supporting SFT, RLVR, On-Policy KD, Guide KD, and mixed training. It is built on OpenRLHF, with the core part being openrlhf-kd, which streamlines the RLVR part and removes components such as the critic, while adding knowledge distillation (KD), Guide-KD, and mixed training with reward, and supports routing by datasource. The SFT part in the root directory provides a concise training entry point, suitable for rapid fine-tuning experiments, supports LoRA, QLoRA, and full-parameter fine-tuning, and can automatically adapt to chat templates.

Read more

LLM-eval-survey

AI developer platforms

LLM-eval-survey is the official GitHub repository of the large language model evaluation survey paper "A Survey on Evaluation of Large Language Models," used to compile and organize papers and resources related to LLM evaluation. The repository organizes papers according to the chapter structure of the survey, covering dimensions such as What to evaluate and Where to evaluate, and continuously collects evaluation literature in directions such as natural language processing, robustness, ethics, bias and trustworthiness, social sciences, natural sciences and engineering, medical applications, and agent applications. The repository also provides links to related projects, such as PromptBench and LLM-eval for large language model robustness evaluation.

Read more

llm-gateway

AI developer platforms

Manifest LLM Gateway is an open-source LLM gateway for AI agents and applications. It can unify API keys, subscriptions, and local models into a single OpenAI-compatible endpoint, so that each query is automatically routed to the appropriate model, avoiding lock-in to a single vendor. Main capabilities include custom routing, full request body logging for both success and error messages, cost tracking and limit notifications, automatic fallback to other models when a query fails, and self-healing of erroneous requests. It supports self-hosting via Docker images and also offers multiple one-click deployment options such as Railway, Render, AWS, GCP, Fly.io, and more. It can connect to more than 300 models, has 35 built-in provider connections, and supports any custom endpoint compatible with OpenAI/Anthropic.

Read more