Dominic789654/awesome-deepseek-harness
A curated list of plugins, skills, MCP servers, patch/profile layers, orchestrators & UIs for DeepSeek Harness (DSH). Visualization · PPT · Coding · Agents · Loops (auto-research) and more. #dsh
Ph.D. student @ HKUST(GZ). Efficient LLM inference, long-context evaluation, KV cache, and agent systems. Builder of awesome-deepseek-harness.
Repositories
Dominic789654/scholar_tracker
Tracks Google Scholar citation counts, h-index, and per-paper statistics on a daily schedule and publishes markdown reports.
Dominic789654/Dominic789654
GitHub profile README for Xiang Liu, summarizing research on efficient LLM inference, long context, KV cache, and agent systems.
Dominic789654/XiangL.github.io
Redirect page that forwards visitors from XiangL.github.io to Xiang Liu's personal homepage.
Dominic789654/models
Placeholder repository for storing model weights; it currently contains only a short README.
Dominic789654/xiangL
Empty placeholder repository; no files have been committed.
Dominic789654/MyArxiv
Personalized arXiv paper digest site built on the MyArxiv template, with config-driven feeds and daily GitHub Actions updates.
Dominic789654/Web-homework
Laravel course assignment implementing a small web application with authentication and book management Blade views.
Dominic789654/tinker2openai-tool
Compatibility helpers for Tinker/Qwen XML-like tool calls and OpenAI tool_calls
Dominic789654/TianChi_Note
Study notes for TianChi competition practice; the repository currently contains only a short README.
Dominic789654/read_note
Reading notes on AI and LLM papers and articles, organized by category code with reusable templates and an index.
Dominic789654/personal-knowledge-base
个人 AI 知识库 · RAG second brain indexing PDFs, Markdown, web pages and GitHub repos for semantic search and cited question answering.
Dominic789654/Paper-List
Reserved for a curated reading list of research papers; the repository is currently empty.
Dominic789654/multica-skills
Public collection of agent skills for research writing, paper review, figure planning, and draw.io diagrams.
Dominic789654/model-case-repros
Small reproducible test cases for OpenAI-compatible chat completion endpoints, plus recorded examples of failure responses.
Dominic789654/longcat-omni-analysis
Technical write-up comparing open-source omni-modal LLMs including LongCat-Flash-Omni, Qwen3-Omni, and MiniCPM-o.
Dominic789654/llm_benchmark_eval
Framework for evaluating LLMs on math and science reasoning benchmarks such as MATH, GSM8K, AIME, and GPQA using xVerify scoring.
Dominic789654/ICR
Notebook solution for the Kaggle ICR Identifying Age-Related Conditions competition, working from the provided health-record CSVs.
Dominic789654/homebrew-kimiquota
Homebrew tap for KimiQuota - macOS menu bar app for Kimi Coding Plan
Dominic789654/football-prediction-skill
An agent skill that turns football betting into evidence-grounded probability judgment. 5-step workflow + 4 data sources. Bilingual.
Dominic789654/EvoHarness.skills
Codex skills for trace-driven harness evolution
Dominic789654/energy-to-token
Project page for: LLM Inference Should Be Evaluated as Energy-to-Token Production (position paper, 2026)
Dominic789654/ell_demo
Minimal demo of the ell library: decorated prompt functions that generate story ideas, write drafts, then pick the best one.
Dominic789654/CS425_Final
Course final project: a 2D roguelike dungeon game for macOS with box-opening levels, monster combat, and scripted game AI.
Dominic789654/claude-code-skills
Claude Code delegation skill - delegate tasks to Claude Code CLI as a subagent
Dominic789654/C4-downloader
Downloads a large C4 corpus subset by paging the Hugging Face datasets-server API and saving the rows to a local JSON file.
Dominic789654/auto-research
Pure-skill autonomous experiment loop: hypothesis → edit → measure → keep wins / revert losses. CLI-grade measurement honesty as protocol rules, no install.
Dominic789654/ai-semiconductor-bottleneck-atlas
Static web deck tracking AI data-center semiconductor supply-chain bottlenecks such as HBM, CoWoS packaging, and 1.6T optics.
Dominic789654/agentic-dataset-builder-ts
TypeScript toolkit with a lossless ATIF and Qwen chat-record codec plus a builder that turns agent session history into parquet datasets.
Dominic789654/agentic-dataset-builder
Scans local Pi and Codex agent session history, labels each record, and merges the kept records into one final parquet dataset.