# BLOOM Projects at AI Tinkerers

> Canonical HTML: https://aitinkerers.org/technologies/bloom?page=4
> Markdown URL: https://aitinkerers.org/technologies/bloom.md?page=4
> Technology record last updated: 2026-02-24T08:39:03Z
> Generated: 2026-08-24T08:41:45Z

A 176-billion parameter open-access multilingual language model built by the BigScience research collective.

BLOOM is the result of a year-long collaboration involving 1,000+ researchers from 70+ countries. It supports 46 natural languages and 13 programming languages: it provides a high-performance alternative to proprietary models. The model was trained on the Jean Zay supercomputer in France using the 1.6-terabyte ROOTS dataset (a massive collection of diverse text sources). By providing full access to its weights and training process, BLOOM enables global developers to build and audit AI tools without the restrictions of closed-door APIs.

- Official technology site: https://huggingface.co/bigscience/bloom
- Public AI Tinkerers demos and talks: 115
- Result page: 4 of 5

## Recent Public Talks and Demos

### [Unlocking Insights from Tabular Data with LLMs](https://toronto.aitinkerers.org/talks/rsvp_EVqtcfX0ujE)

This application demonstrates how we can leverage Large Language Models (LLMs) to extract valuable insights directly from your company's tabular data. Unlike unstructured data (PDFs, Docs etc.) that require Retrieval-Augmented Generation (RAG) techniques, this approach allows LLMs to interact with structured data in a more efficient way. Here's how it works: Natural Language Understanding with LLM: The user submits a question in plain text. The LLM parses this question to understand its intent and the relevant data points needed. SQL Query Generation: Based on the understanding, the LLM generates a corresponding SQL query to retrieve the desired information from your company's data warehouse. Data Retrieval and Summarization: The generated SQL query is executed, fetching the relevant data points. Finally, the LLM summarizes the retrieved data in a clear and concise manner, providing a comprehensive answer to the user's initial question.

- Event context: AI Tinkerers - Toronto - June 2024 Meetup — 2024-06-27 — Toronto
- Public talk page: https://toronto.aitinkerers.org/talks/rsvp_EVqtcfX0ujE

### [OpenSesame Demo](https://toronto.aitinkerers.org/talks/rsvp_7C5CZ6quT8M)

OpenSesame is a plug-and-play API that detects AI hallucinations and improves response accuracy for companies using LLMs. Our platform leverages context-based information to detect hallucinations in LLMs with precision. By ensuring accurate and error-free text generation, we provide a reliable solution for enhancing AI outcomes.

- Event context: AI Tinkerers - Toronto - June 2024 Meetup — 2024-06-27 — Toronto
- Public talk page: https://toronto.aitinkerers.org/talks/rsvp_7C5CZ6quT8M

### [AI-powered Legal Contract Analysis POC](https://kuala-lumpur.aitinkerers.org/talks/rsvp_SRNn5lVdOsI)

We are using LLMs to help do legal contract analysis in the following areas: 1. Clause Identification (and custom Key clause identification) 2. Clause Recommendation 3. RAG over Contract

- Event context: AI Tinkerers Kuala Lumpur - June 27th — 2024-06-27 — Kuala Lumpur
- Public talk page: https://kuala-lumpur.aitinkerers.org/talks/rsvp_SRNn5lVdOsI

### [Anatomy of a Thinking Machine](https://la.aitinkerers.org/talks/rsvp_MjNJg6eHsLw)

We all know that AI has a hardware problem. - What happens at a hardware level during inference? - What exactly are all these tools in the inference ecosystems from Nvidia, AMD...? - An early preview of Cortex, an open source tool that runs LLMs across multiple platforms Here's my cofounder Dan doing a similar talk, but I'm planning for this demo to be **shorter & purely technical**: https://www.youtube.com/watch?v=orcPcUzSbOw&ab_channel=HackerHouseTW

- Event context: June 25th - LA AI Tinkerers Meetup & Demos — 2024-06-26 — Los Angeles
- Public talk page: https://la.aitinkerers.org/talks/rsvp_MjNJg6eHsLw

### [Scripter Studio -- Founders University Cohort 8](https://la.aitinkerers.org/talks/rsvp_tSKu0qwQ1JA)

An Ai powered tool that assists the Hollywood industry in script scheduling, coverage, analysis

- Event context: June 25th - LA AI Tinkerers Meetup & Demos — 2024-06-26 — Los Angeles
- Public talk page: https://la.aitinkerers.org/talks/rsvp_tSKu0qwQ1JA

### [LLM-powered Knowledge Graph Generation](https://boston.aitinkerers.org/talks/rsvp_DVbe4XH70-Q)

We'll be running code to test how well LLMs can build knowledge graphs from scratch starting from a single node.

- Event context: June 2024 Meetup at C10 Labs — 2024-06-10 — Boston
- Public talk page: https://boston.aitinkerers.org/talks/rsvp_DVbe4XH70-Q

### [Pick-Em's Bot](https://seattle.aitinkerers.org/talks/rsvp_H6MqDp0e2zM)

This is an LLM agent built that scrapes sports articles and team stats to predict winners for ESPN's Pick-Em's game. For the 2023 NFL season, this agent won my local group's pick-em's competition and came in the top 15% across all of ESPN's users (~one million users played). The top player across all of ESPN for 2023 correctly chose 195 games (~71%). This agent correctly chose 163 (~60%).

- Event context: AI Tinkerers Seattle - June 2024 — 2024-06-07 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_H6MqDp0e2zM

### [LLM-powered Knowledge Graph Generation](https://nyc.aitinkerers.org/talks/rsvp_hauOLk3d7qc)

Generalized LLMs have made tasks like entity tagging and relationship extraction from unstructured text much easier. This enables developers to create structured networks that are important for relational insight out of unstructured data.

- Event context: Omakase Founder Dinner | NY #TechWeek — 2024-06-04 — New York City
- Public talk page: https://nyc.aitinkerers.org/talks/rsvp_hauOLk3d7qc

### [Reliable AI Agents](https://nyc.aitinkerers.org/talks/rsvp_TE3Go21sxYY)

AI agent demos based on LLMs are so common, but as Tinkerers we know they have many pitfalls around accuracy and hallucination. If we alter the prompt just slightly, the output completely changes! Grounded in years of dialogue systems research, I will be presenting the latest features in an agent built using classical approaches to Conversational AI. Rather than LLMs+RAG, this agent has an special NLU module (natural language understanding) that explicitly predicts and tracks specific named entities in the conversation, allowing for unprecedented control and reliability.

- Event context: AI Tinkerers NYC May 2024 Meetup — 2024-05-22 — New York City
- Public talk page: https://nyc.aitinkerers.org/talks/rsvp_TE3Go21sxYY

### [Twitter '95 (and Fireside Chat)](https://nyc.aitinkerers.org/talks/rsvp_wlGwEKyF4lw)

This project imagines what Twitter would be like if it had been invented in 1995, using LLMs to mimic social media participants and historical information to inject grounding.

- Event context: AI Tinkerers NYC May 2024 Meetup — 2024-05-22 — New York City
- Public talk page: https://nyc.aitinkerers.org/talks/rsvp_wlGwEKyF4lw

### [LLMs and Time Travel](https://la.aitinkerers.org/talks/rsvp_40t5zXBsD0U)

Demo of the synthasaizer - a new open source package to control agents and send them to different time points for market analysis and strategic planning.

- Event context: May 21st - LA AI Tinkerers Meetup & Demos — 2024-05-22 — Los Angeles
- Public talk page: https://la.aitinkerers.org/talks/rsvp_40t5zXBsD0U

### [Beyond Linear Chat - AI UX Exploration in Obsidian](https://denver-boulder.aitinkerers.org/talks/rsvp_pAVIn4Du3Wo)

ChatGPT and similar forces you to have linear conversation. You send a message, the assistant responds. But conversations in real life are multi-dimensional. You go down a conversation branch for awhile, then circle back to another topic, or explore a whole new area. And then maybe you synthesize across the different conversation branches. By using the Obsidian's (note-taking app) canvas we can enable this multi-dimensional conversational.

- Event context: AI Tinkerers Denver - May Meetup — 2024-05-14 — Denver
- Public talk page: https://denver-boulder.aitinkerers.org/talks/rsvp_pAVIn4Du3Wo

### [Extracting Meanings from Qualitative Data using LLMs](https://zurich.aitinkerers.org/talks/rsvp_txF5s-omAfI)

Qualitative research is one of the most popular techniques to gather meaningful data in social science, psychology, marketing, and even medical studies. However, there are currently no methods that analyze vast collections of documents (interviews, focus groups, reviews, etc.) dedicated to discovering people's thoughts and opinions. Most NLP techniques offer a 'quantitative' approach to analyzing texts - for instance, calculating word frequencies and building a word cloud, Topic Modeling, Semantic Analysis, and others. This project aims to compare two paradigms: analyzing texts VS analyzing meanings. For example the sentence 'I feel I can't stop using my fitness tracker, it is controlling my whole life' can be denoted as 'addiction'. Interpretations are inherently subjective, varying among individuals (another person could say this sentence is about 'control', not 'addiction') and language models produce different results as well. To illustrate these differences, I fine-tuned 10 LLMs on a curated dataset and demonstrated how their interpretations diverge from those derived through Topic Modeling.

- Event context: AI Tinkerers Zurich - May 8 — 2024-05-08 — Zürich
- Public talk page: https://zurich.aitinkerers.org/talks/rsvp_txF5s-omAfI

### [Vocal Docs - A smart document editor you can talk to](https://seattle.aitinkerers.org/talks/rsvp_cP8BTNWsLeM)

A smart document editor that you can talk to, both for dictation but also editing.

- Event context: AI Tinkerers Seattle - April 2024 Meetup — 2024-04-26 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_cP8BTNWsLeM

### [RAG Copilot: Big Data-powered solutions for rapid Information extraction](https://medellin.aitinkerers.org/talks/rsvp_1rdidJ3owc8)

Solution for Automatic Content Extraction from Large Data Volumes Using a RAG (Retrieval Augmented Generation) Model and AWS Serverless Architecture with a MapReduce Approach

- Event context: Lanzamiento de AI Tinkerers #1 — 2024-04-25 — Medellín
- Public talk page: https://medellin.aitinkerers.org/talks/rsvp_1rdidJ3owc8

### [CrustyCrab: An Experimental LLM-based C-to-Rust Translator](https://nyc.aitinkerers.org/talks/rsvp_9pD1IUpTTMI)

CrustyCrab is an experimental translator that uses LLMs to translate C to idiomatic Rust. Converting legacy, unsafe C code to memory-safe Rust is becoming a higher priority at many organizations. Recently, the White House recommended that all software should be written in memory safe languages. Unfortunately, manual translation of legacy codebases is a time-intensive, high-skill affair. Automated translators exist to expedite this process, but current algorithmic translators do not create idiomatic, safe Rust code. Instead, they simply emulate the original C via unsafe Rust. CrustyCrab aims to work around the limitations of algorithmic translation by using LLMs to translate C to safe, idiomatic Rust. In this talk we’ll describe how CrustyCrab works, our current results, and our plans to address current shortcomings.

- Event context: AI Tinkerers NYC April 2024 Meetup — 2024-04-24 — New York City
- Public talk page: https://nyc.aitinkerers.org/talks/rsvp_9pD1IUpTTMI

### [Unnamed language learning app](https://boston.aitinkerers.org/talks/rsvp_e0Nz2fQ-j28)

App prototype to use llms for language learning

- Event context: March 2024 Meetup at C10 Labs — 2024-03-25 — Boston
- Public talk page: https://boston.aitinkerers.org/talks/rsvp_e0Nz2fQ-j28

### [AI Litigation Chess: presenting Libra's chain of thought algorithm to analyse and attack legal arguments](https://berlin.aitinkerers.org/talks/rsvp_yfrs6ExVPd8)

Viktor von Essen, Bo Tranberg & Lukas Kemes (team Libra), will discuss the challenges and opportunities of building AI applications in the legal space, present first use-cases and progress.

- Event context: AI Tinkerers Berlin - March 21 — 2024-03-21 — Berlin
- Public talk page: https://berlin.aitinkerers.org/talks/rsvp_yfrs6ExVPd8

### [Viewpoint.AI accelerated augmentation of individual and complex group decisions, by 10-1000x, Fully transparent, continuously curated by humans, delivering optimal results fro Finech, Legal, Education, Aerospace, IT, Governement and more.](https://ee.dream.page/talks/rsvp_5DqayfQPAfg)

Viewpoint AI enables groups and organizations to make complex decisions up to 100 times faster with optimal, transparent results. We apply multiple AI techniques to build and utilize verified AI digital representations of human, individual, and group decision-thinking while capturing, preserving, and organizing expert and institutional knowledge for current and future collaboration. Our patent-pending AI-generated decisions are based on innovations with human knowledge modeling (HKM). We make best use of AI technologies and patterns including, LLMs, NLPs, Neural Nets, and Machine Learning We designed our system to accommodate specific industries and processes. We currently have industry templates and models for angel and venture capital, human resources, legal, and higher learning admissions. Current customer examples include: Should we invest in this company? Should we hire this person? Should we enroll these people into our university? And many more complex decisions are available. We have many AI patents pending:

- Event context: Bellevue + Eastside Entrepreneurs / March 7, 2024 — 2024-03-08 — Eastside Entrepreneurs
- Public talk page: https://ee.dream.page/talks/rsvp_5DqayfQPAfg

### [Automated Social Science: A Structural Causal Model-Based APproach](https://boston.aitinkerers.org/talks/rsvp_pl4pDy5b4XA)

We present an approach for automatically generating and testing, in silico, social scientific hypotheses. This automation is made possible with recent advances in large language models (LLM), but the key feature of the approach is the use of structural causal models (SCM). SCMs provide a language to state hypotheses, a blueprint for constructing LLM-based agents, an experimental design, and a plan for data analysis. The fitted SCM becomes an object available for prediction or the automated planning of follow-on experiments. We demonstrate the approach with several scenarios: a negotiation, a bail hearing, a job interview, and an auction. In each case, causal relationships are proposed and tested, finding evidence for some and not others. In the auction experiment, we show that the in silico simulation results closely match the predictions of auction theory, but elicited predictions of the clearing prices from the LLM are highly inaccurate. However, the LLM's clearing price predictions are dramatically improved if the model can condition on the fitted SCM. When given a proposed SCM for one of the scenarios, the LLM is good at predicting the signs of estimated effects, but it cannot reliably predict the magnitudes of those effects. This suggests that explicit social simulation gives the model insight not available purely through direct elicitation. In short, the LLMs know more than they can (immediately) tell.

- Event context: February 2024 Tinkerers Unite! — 2024-02-26 — Boston
- Public talk page: https://boston.aitinkerers.org/talks/rsvp_pl4pDy5b4XA

### [Gotta Deploy ‘Em All! Applying LLMs for Pokemon Battle Calculations](https://chicago.aitinkerers.org/talks/rsvp_FfF7KF6XUOw)

I'll present my personal project, which involves building a command line-like assistant for competitive Pokemon players. I built a web app that accepts natural language queries, and returns relevant battle calculations for Pokemon players about 5x faster than the current best calculation software. Very often when playing online, players will need to estimate the probability that a Pokemon will knockout, two-hit-ko, or outspeed the opposing pokémons with incomplete information about the game state. Online calculators offer this capability, but you have to navigate airplane dashboard-like interfaces to use them. Since players already describe these game states in a specific lingo, I decided to create an app that can directly translate this natural language query to the calculation!

- Event context: AI Tinkerers Chicago February Meetup — 2024-02-20 — Chicago
- Public talk page: https://chicago.aitinkerers.org/talks/rsvp_FfF7KF6XUOw

### [Local LLM Social Media Filter](https://chicago.aitinkerers.org/talks/rsvp_ZKE5CrbzojE)

AI Filter is a Chrome extension that uses a local language model to filter your social media feeds (currently, only Twitter / X) according to your instructions. For instance, you can say: Hide all tweets, except for tweets about machine learning (ML), artificial intelligence (AI) and large language models (LLMs). or: By default, show all tweets Do not show any tweets related to cryptocurrencies, blockchain, Bitcoin, Ethereum or related projects. AI Filter will follow your instructions, hiding the appropriate tweets in the background as you browse Twitter.

- Event context: AI Tinkerers Chicago February Meetup — 2024-02-20 — Chicago
- Public talk page: https://chicago.aitinkerers.org/talks/rsvp_ZKE5CrbzojE

### [ClaimValidator](https://la.aitinkerers.org/talks/rsvp_UL1dwuNxlGg)

A browser extension that uses LLMs to *increase* trust on the internet. When someone provides a source for their claims in a Reddit comment, we can quickly verify whether or not the claims are really substantiated by the source or not in a report that is provided in-line.

- Event context: Feb 8th - AI Tinkerers Meetup — 2024-02-09 — Los Angeles
- Public talk page: https://la.aitinkerers.org/talks/rsvp_UL1dwuNxlGg

### [Extraction: Making Using Tools With OpenAI Clean And Simple](https://la.aitinkerers.org/talks/rsvp_Sq2DRFV4644)

We recently implemented some functionality that enabled you to define tools as Pydantic models so that you get validation. Additionally, you can use functions directly as tools with no additional work if documented properly. Furthermore, you can extract information directly into a Pydantic model, built using this functionality

- Event context: Feb 8th - AI Tinkerers Meetup — 2024-02-09 — Los Angeles
- Public talk page: https://la.aitinkerers.org/talks/rsvp_Sq2DRFV4644

## Related Technologies

- [BERT](https://aitinkerers.org/technologies/bert) ([Markdown](https://aitinkerers.org/technologies/bert.md)) — 179 public demos
- [GPT-3](https://aitinkerers.org/technologies/gpt-3) ([Markdown](https://aitinkerers.org/technologies/gpt-3.md)) — 191 public demos
- [GPT-4](https://aitinkerers.org/technologies/gpt-4) ([Markdown](https://aitinkerers.org/technologies/gpt-4.md)) — 529 public demos
- [Llama-2](https://aitinkerers.org/technologies/llama-2) ([Markdown](https://aitinkerers.org/technologies/llama-2.md)) — 227 public demos
- [PaLM 2](https://aitinkerers.org/technologies/palm-2) ([Markdown](https://aitinkerers.org/technologies/palm-2.md)) — 116 public demos
- [RoBERTa](https://aitinkerers.org/technologies/roberta) ([Markdown](https://aitinkerers.org/technologies/roberta.md)) — 118 public demos
- [RAG](https://aitinkerers.org/technologies/rag) ([Markdown](https://aitinkerers.org/technologies/rag.md)) — 146 public demos
- [scikit-learn](https://aitinkerers.org/technologies/scikit-learn) ([Markdown](https://aitinkerers.org/technologies/scikit-learn.md)) — 83 public demos
- [TensorFlow](https://aitinkerers.org/technologies/tensorflow) ([Markdown](https://aitinkerers.org/technologies/tensorflow.md)) — 90 public demos
- [Keras](https://aitinkerers.org/technologies/keras) ([Markdown](https://aitinkerers.org/technologies/keras.md)) — 74 public demos
- [ONNX](https://aitinkerers.org/technologies/onnx) ([Markdown](https://aitinkerers.org/technologies/onnx.md)) — 83 public demos
- [PyTorch](https://aitinkerers.org/technologies/pytorch) ([Markdown](https://aitinkerers.org/technologies/pytorch.md)) — 273 public demos
- [Python](https://aitinkerers.org/technologies/python) ([Markdown](https://aitinkerers.org/technologies/python.md)) — 654 public demos
- [Generative AI](https://aitinkerers.org/technologies/generative-ai) ([Markdown](https://aitinkerers.org/technologies/generative-ai.md)) — 45 public demos
- [Large Language Models](https://aitinkerers.org/technologies/large-language-models) ([Markdown](https://aitinkerers.org/technologies/large-language-models.md)) — 8 public demos
- [Prompt Engineering](https://aitinkerers.org/technologies/prompt-engineering) ([Markdown](https://aitinkerers.org/technologies/prompt-engineering.md)) — 28 public demos
- [Fine-tuning](https://aitinkerers.org/technologies/fine-tuning) ([Markdown](https://aitinkerers.org/technologies/fine-tuning.md)) — 20 public demos
- [AI agents](https://aitinkerers.org/technologies/ai-agents) ([Markdown](https://aitinkerers.org/technologies/ai-agents.md)) — 35 public demos

## More Results

- Previous: https://aitinkerers.org/technologies/bloom.md?page=3
- Next: https://aitinkerers.org/technologies/bloom.md?page=5