Chatbot2023
Turns text, images, voice, files, and live context into conversational help across web and mobile. Its edge is tight access to Google Search, Android, Workspace, and multimodal Gemini models; the tradeoff is ecosystem lock-in and uneven reliability.
Introduced the two-stage recipe behind the GPT lineage: unsupervised generative pre-training on unlabeled text, then supervised fine-tuning per task. A single 12-layer Transformer decoder beat bespoke architectures on 9 of 12 NLP benchmarks.
AI Model2018
Turns model definitions into a shared layer across training and inference stacks, covering text, vision, audio, video, and multimodal models. Pipelines, Trainer, and generation APIs make pretrained models usable without locking teams to one framework.
AI Infra2021
Drop-in transformer building blocks with custom CUDA kernels: memory-efficient exact attention (up to ~10x faster), block-sparse attention, fused softmax/layernorm/SwiGLU ops. Cuts VRAM and speeds up diffusion and LLM training on Nvidia GPUs.
AI Model2022
Transformer-based foundation model for tabular data that provides pre-trained checkpoints for fast classification and regression, with GPU-accelerated local inference and an optional cloud client. Best suited for small-to-medium datasets (~≤50k rows).
AI Train2023
Connects a frozen vision encoder to a language model via visual instruction tuning, yielding an open multimodal assistant that follows image-grounded instructions. Released checkpoints span 7B-34B and approach GPT-4V on vision-language benchmarks.
AI Train2023
Fine-tunes and deploys 600+ LLMs and 400+ multimodal models in one framework, with SFT, pretraining, RLHF (DPO, PPO, GRPO), and lightweight methods like LoRA and QLoRA. Adds Megatron parallelism, vLLM/SGLang/LMDeploy inference, and a training web UI.
Chatbot2023
Runs a consumer AI chat interface backed by DeepSeek's large language models, with text, code, table, file, app, and API workflows. Its main appeal is strong reasoning access at unusually low cost.
AI Infra2023
A selective State Space Model architecture and PyTorch implementation for linear-time sequence modeling. Hardware-aware, designed for information-dense tasks (e.g. language modeling), with pretrained weights on Hugging Face; requires CUDA-enabled PyTorch.