Jimin Huang

[Jimin Huang @JiminH_AI]

📷 Excited to introduce PIXIU - the first open-source financial instruction dataset, benchmark, and fine-tuned LLMs! This is a big step forward for #FinancialAI. Check out our #arXiv paper (https://t.co/SvNl6Im5r1) and GitHub repo (https://t.co/S0gaG2c5SP)! #AI #NLP #FinTech

[Jimin Huang @JiminH_AI]

@PaalMind @alejandroll10 Proud to welcome our newest Sustaining Partner to https://t.co/GSBQ91JkzA!

let’s build the future of financial AI together.

[PAAL AI 🧠🤖 @PaalMind]

We’ve joined the https://t.co/hcLrujIWR9 community as a Sustaining Partner — a global research group backed by talent from Harvard, Columbia, and University of Florida. 🤝

With @jiminh_ai and @alejandroll10 leading the charge, we're blending academic rigor with real-world performance — from reasoning benchmarks to multimodal agents.

Keep a lookout for the AI trading agent we’ll be releasing together in Q2 — built for transparency, performance, and real-world utility.

The future of financial AI is open. Let’s build it together.

[Alejandro Lopez-Lira @alejandroll10]

We are happy to announce the launch of the FinNLP-AgentScen at IJCAI-2024, FinLLM, which includes three subtasks tailored to effectively and holistically address diverse financial challenges:

- Financial Text Classification: Designing LLMs to categorize financial texts as "Claim" or "Premise."

- Financial News Summarization: Creating LLMs to generate concise summaries of financial news, testing LLMs' ability to produce coherent narratives.

- Financial Stock Trading based on Agents: Evaluating LLMs on sophisticated trading decisions, using a mix of stock and ETF data informed by the FinMem agent.

Key Dates for the Challenge Phase:

- Registration Open: April 29th, 2024

- Training Set Release: May 1st, 2024

- Test Set Release & Submission Deadline: May 28th, 2024

- Model Cheating Detection: May 29th, 2024

- Results Release: June 1st, 2024

Paper Phase:

- Submissions Due: June 15th, 2024

- Notification: June 20th, 2024

- Camera-Ready Version Due: June 25th, 2024

For detailed information, including how to get started, please visit our [challenge webpage](https://t.co/0w9Q09jAxP).

Please reach out via email at [email protected] with any questions or suggestions.

We eagerly anticipate your active participation and innovative contributions!

[Ethan Mollick @emollick]

It is very possible the highest quality training data for AI is a freely available pirated dataset that contains half a petabyte of copyrighted books and articles.

It is extremely unlikely that any commercial LLM maker would use it, but governments & open source might decide to.

[Alejandro Lopez-Lira @alejandroll10]

My first paper in #NeurIPS2023! We produce open-source benchmarks and training data for LLMs in finance tasks, and even a fine tuned model!

[Jimin Huang @JiminH_AI]

🎉 "PIXIU: A Benchmark & LLM for Finance" is now accepted by #NeurIPS2023! Dive in:
🔗Paper/Datasets/Code: https://t.co/0KKpKB5A40
Thanks to Qianqian Xie, Weiguang Han, @alejandroll10 & team! 🙌 #FinancialNLP #AI #LLMs #fintech

[Aris K. @ArisJanus9]

First milestone just reached! Thank you everyone for your support! Defining our next one soon.. https://t.co/2AlAYSLgtC
#machinelearning #deeplearning #ArtificialIntelligence #opensource #PyTorch

[Jimin Huang @JiminH_AI]

Wrapped up #KDD2023 with our paper on hierarchical RL in pair trading. Energized by the synergy of finance & data science, especially with financial LLMs & #Pixiu. Looking forward to #KDD2024!
#DataScience #PairTrading

[Jimin Huang @JiminH_AI]

Currently at #KDD2023, Long Beach. Let's chat about AI's role in finance and explore FinLLM's Pixiu project. Open for conversations. #AI #Finance #FinLLM #Pixiu

[SIGKDD 2026 @kdd_news]

The hottest topic of the premier conference! Please attend the LLM day at KDD-2023 next week! August 8, Los Angeles. https://t.co/1KE0ISROKm

[Alejandro Lopez-Lira @alejandroll10]

💹New results in the space of open source financial LLMs:

💻We added Llama 2 to our Financial Natural Language Understanding and Prediction Evaluation Benchmark (#FLARE) Leaderboard

👨‍💻 #FinMA looking well and we're currently improving it! @JiminH_AI

https://t.co/ONtJmCRSJ4

[SIGKDD 2026 @kdd_news]

📢 Just released: Paper schedules for #KDD2023! Plan ahead and don't miss a beat.
🔍Research track: https://t.co/UF510ejMob
📈ADS Track: https://t.co/DwhKWTA0Im
Dive into the conference hype and get a sneak peek by watching their promo videos on YouTube: https://t.co/shs87MK9N6

[Alejandro Lopez-Lira @alejandroll10]

PIXIU update: FLARE benchmarks now evaluate all large language models in #Huggingface, and also #GPT4, for financial #NLP and forecasting. Empower your #FinTech with our open-source FinMA and FIT dataset! #AI #MachineLearning #LLM
@JiminH_AI

[Jimin Huang @JiminH_AI]

🚀 Quick update from the #PIXIU project! We've released the full version of our financial LLM, FinMA-7B-full. Plus, we've added Falcon-7B fine-tuned results on FLARE. Check out our leaderboard: https://t.co/3TdC0uvomt and the new model: https://t.co/6wFGgFRGoB #AI #NLP #Finance

[Agam Shah @shahagam4]

1. "PIXIU: A Large Language Model, Instruction Data and Evaluation Benchmark for Finance" by @JiminH_AI, @alejandroll10, and others

Paper on Arxiv: https://t.co/cZv80iZtEt GitHub Repo: https://t.co/whZihncKvx

[Jimin Huang @JiminH_AI]

@shahagam4 You're welcome! I'm glad I could help. If you have any more questions or need further clarification, feel free to ask.

[Jimin Huang @JiminH_AI]

📷 Excited to introduce PIXIU - the first open-source financial instruction dataset, benchmark, and fine-tuned LLMs! This is a big step forward for #FinancialAI. Check out our #arXiv paper (https://t.co/SvNl6Im5r1) and GitHub repo (https://t.co/S0gaG2c5SP)! #AI #NLP #FinTech

[Jimin Huang @JiminH_AI]

@shahagam4 Appreciate your interest! We ensured a fair comparison of FinMA-7B with models like ChatGPT & GPT-4 by fine-tuning it strictly on NLP instruction data. It is comparable without financial prediction data. Yet future work is needed on more NLP and prediction tasks for evaluation.

[Jimin Huang @JiminH_AI]

Huge thanks to the contributors behind PIXIU and the FinMA models: Qianqian Xie, Weiguang Han, Xiao Zhang, Yanzhao Lai, Min Peng, and @alejandroll10. #PIXIU #AI #NLP #FinTech

[Jimin Huang @JiminH_AI]

Let's take a closer look at how our FinMA models outperform others. In the FPB dataset, FinMA 30B achieves an accuracy of 0.87 and an F1 score of 0.88, surpassing models like GPT-4 and Bloomberg GPT. #PIXIU #FinancialAI #NLP

[Jimin Huang @JiminH_AI]

📷 Excited to introduce PIXIU - the first open-source financial instruction dataset, benchmark, and fine-tuned LLMs! This is a big step forward for #FinancialAI. Check out our #arXiv paper (https://t.co/SvNl6Im5r1) and GitHub repo (https://t.co/S0gaG2c5SP)! #AI #NLP #FinTech

[Jimin Huang @JiminH_AI]

Find more exciting results on our leaderboard https://t.co/3TdC0uvomt

[Jimin Huang @JiminH_AI]

In the Headline dataset, our FinMA 7B model hit an impressive AvgF1 score of 0.98, showing the power of specialized financial AI over models like GPT-4 and Bloomberg GPT. #PIXIU #AI #NLP #FinTech