@MarcHadfield profile picture
Marc Hadfield @MarcHadfield
Joined October 2007
  • Tweets 2,548
  • Following 392
  • Followers 228
46 Photos and videos
Pinned Tweet
Marc Hadfield@MarcHadfield
21 Jan 2025
Ensemble Reasoning with the new Deepseek R1 and Qwen QwQ Reasoning LLMs using Knowledge Graph Queries, Code Execution, and Assisting LLMs. Drive A.I. Agents to be faster, smarter, and more efficient. #llm #reasoning #agents #ai blog.vital.ai/2025/01/21/ens…

Ensemble Reasoning with the Deepseek R1 and Qwen QwQ LLMs

Note:Background information about Ensemble Reasoning can be found in my previous article: Deepseek released the R1 LLM today, January 20th, and the Qwen team from Alibaba released the QwQ Preview L…

blog.vital.ai
1
131
Marc Hadfield retweeted
Bowen Jin
@BowenJin13
28 Feb 2025
🚀 Introducing 𝗦𝗲𝗮𝗿𝗰𝗵-𝗥𝟭 – the first 𝗿𝗲𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻 𝗼𝗳 𝗗𝗲𝗲𝗽𝘀𝗲𝗲𝗸-𝗥𝟭 (𝘇𝗲𝗿𝗼) for training reasoning and search-augmented LLM agents with reinforcement learning! This is a step towards training an 𝗼𝗽𝗲𝗻-𝘀𝗼𝘂𝗿𝗰𝗲 𝗢𝗽𝗲𝗻𝗔𝗜 “𝗗𝗲𝗲𝗽 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵” via RL. Our 𝟯𝗕 𝗯𝗮𝘀𝗲 𝗟𝗟𝗠𝘀—including not just 𝗤𝘄𝗲𝗻 𝟮.𝟱 but also 𝗟𝗹𝗮𝗺𝗮 𝟯.𝟮—learn to 𝗿𝗲𝗮𝘀𝗼𝗻 and 𝗰𝗮𝗹𝗹 𝘀𝗲𝗮𝗿𝗰𝗵 𝗲𝗻𝗴𝗶𝗻𝗲𝘀 all on their own! Everything will be 𝗳𝘂𝗹𝗹𝘆 𝗼𝗽𝗲𝗻 𝘀𝗼𝘂𝗿𝗰𝗲. Stay tuned! Code: github.com/PeterGriffinJin/S… Experimental logs: wandb.ai/peterjin/Search-R1-… #R1 #deepresearch #deepseek

GitHub - PeterGriffinJin/Search-R1: Search-R1: An Efficient, Scalable RL Training Framework for...

Search-R1: An Efficient, Scalable RL Training Framework for Reasoning & Search Engine Calling interleaved LLM based on veRL - PeterGriffinJin/Search-R1

github.com
42
316
2,530
364,190
Impressive! #ai #agent #voice #llm sesame.com/research/crossing…
49
KGraphLang: Knowledge Graph Query Language for Reasoning LLMs. Enable reasoning LLMs to query knowledge graphs at inference (test) time, dramatically improving A.I. Agent performance by eliminating the latency of tool calls. #ai #agent blog.vital.ai/2025/03/11/kgr…

KGraphLang: Knowledge Graph Query Language for Reasoning LLMs

KGraphLang is a knowledge graph query language designed for Ensemble Reasoning and Reasoning LLMs. Reasoning LLMs are trained to split a request into steps and then consider and follow each step to…

blog.vital.ai
48
Great rundown of various concerns, security and otherwise, of the DeepSeek v3 and R1 models: #ai #llm #r1 #deepseek hiddenlayer.com/innovation-h…

DeepSh*t: Exposing the Security Risks of DeepSeek-R1

DeepSeek-R1 sets new benchmarks for open-weight models, but what are the security risks? Understand the security implications before deploying.

hiddenlayer.com
53
Developing Agents using Ensemble Reasoning and OpenSource Reasoning LLMs like Qwq and Phi-4. blog.vital.ai/2025/01/13/age… #LLM #Agent #AI #KnowledgeGraph

Agents and Ensemble Reasoning

The recent release of open source reasoning models such as QwQ (Qwen) and Phi-4 (Microsoft) have opened up new possibilities in Agents to reason via a methodology that can be called: “Ensembl…

blog.vital.ai
1
98
In case you need a blatant example of LLM censorship, here is deepseek v3 with "Who is the leader of China?". As a high-scoring and less expensive option, how will the community balance commercial interests and fidelity? #LLM #AI
0:16
1
254
Can we separate reasoning abilities from knowledge in models to avoid having to make this tradeoff? Large Concept Models may be a step in this direction.
48
#justicefortoshi #acompleteunknown @bobdylan @RealChalamet merrillmarkoe.substack.com/p…
73
Marc Hadfield retweeted
Fermat's Library
@fermatslibrary
24 Oct 2024
Marvin Minsky recalls how John Nash effortlessly solved his PhD problem
1:48
34
406
2,491
271,272
Marc Hadfield retweeted
Brooklyn Botanic Garden@BrooklynBotanic
24 Sep 2024
While the proposal is moving in the right direction, additional modifications are necessary, specifically a decrease from 15-degrees to 10-degrees slope, to prevent serious shadow impacts to BBG’s most vulnerable collections.
1
1
2
374
Marc Hadfield retweeted
Giliam de Carpentier@decarpentier_nl
10 Sep 2024
By popular request, I've now open sourced the exact dimensions of the evolved Carpentopod's linkage system. So feel free to use it for your own projects! decarpentier.nl/carpentopod
57
492
3,484
154,964
Marc Hadfield retweeted
Brooklyn Botanic Garden@BrooklynBotanic
2 Aug 2024
Attention Fight for Sunlight supporters! ☀️ @NYCPlanning will hold a public hearing on August 7 to evaluate the rezoning application for 962–972 Franklin Avenue. We’re once again asking supporters to join us in opposing this rezoning. bbg.org/support/save_the_sun…

Protect Brooklyn’s Garden: Join the Fight for Sunlight - Brooklyn Botanic Garden

A proposed compromise would update zoning to preserve sunlight to the Garden’s vital resources and support affordable workforce housing.

bbg.org
1
17
25
3,028
Marc Hadfield retweeted
Brooklyn Botanic Garden@BrooklynBotanic
26 Jul 2024
Thank you @BKBPReynoso for your support of BBG, a place for ALL Brooklynites.
Antonio Reynoso@BKBPReynoso
26 Jul 2024
Today I disapproved 962-972 Franklin Ave. @BrooklynBotanic is sacred, and this one-of-a-kind community resource deserves one-of-a-kind consideration. I will not support any development here that will create additional shadow impacts. This proposal fails to meet that standard.
3
6
35
2,754
Marc Hadfield retweeted
Brooklyn Botanic Garden@BrooklynBotanic
10 Jun 2024
BBG’s Fight for Sunlight is back. ☀️ Once again, we are asking for community support in opposing a rezoning that would have devastating impacts on our plant collections. 🧵👇bbg.org/support/save_the_sun…

Protect Brooklyn’s Garden: Join the Fight for Sunlight - Brooklyn Botanic Garden

A proposed compromise would update zoning to preserve sunlight to the Garden’s vital resources and support affordable workforce housing.

bbg.org
15
12
40
11,328
Combining logical reasoning with LLMs with Neuro Symbolic AI and Defeasible Reasoning (with a Python code example). blog.vital.ai/2024/04/05/rea… #llm #logic #reasoning #chatgpt #neurosymbolic #ai

Reasoning, LLMs, Neuro-Symbolic AI, and Defeasible Logic (with Python Example)

I recently had the above interaction with ChatGPT (GPT-4). I asked: There is a standard 3 color traffic light that is displaying a color. The color is not green or blue. What color or colors could …

blog.vital.ai
70
Running Whisper Speech-to-Text Model in the Browser blog.vital.ai/2024/04/03/run…

Running Whisper Speech-to-Text Model in the Browser

In Chat.ai, we’re looking to improve voice access to artificial intelligence. Converting Speech-to-Text is a critical component of interacting with people. Once speech is converted to text it…

blog.vital.ai
42
Marc Hadfield retweeted
Brooklyn Botanic Garden@BrooklynBotanic
12 Mar 2024
CherryWatch 2024 is officially live! 🌸🚨 The first cherry blossoms have been spotted on three trees in the garden: ‘Okame’ in the Cherry Cultivar area, ‘Jugatsu-zakura’ on Cherry Walk, and ‘Fudan-zakura’ in the Japanese Hill-and-Pond Garden. More info: bbg.org/collections/cherries
A map of Brooklyn Botanic garden showing several garden areas. The map shows many kinds of cherry trees, stylized as dots. The vast vast majority aren't blooming yet, as indicated by the key on the bottom of the graphic.

ALT A map of Brooklyn Botanic garden showing several garden areas. The map shows many kinds of cherry trees, stylized as dots. The vast vast majority aren't blooming yet, as indicated by the key on the bottom of the graphic.

1
53
301
80,620
Marc Hadfield retweeted
Kareem Carr, Ph.D.
@kareem_carr
6 Feb 2024
This is the greatest randomized controlled trial joke ever told
56
994
9,139
967,581
Just made public AgentShop.ai Search @OpenAI GPTs to find fun and useful GPTs to try on ChatGPT. Built with: @weaviate_io Virtuoso @OpenLink @awscloud @ChatDotAI blog.vital.ai/2023/12/04/sea… #gpt
2
4
723
Load more