Interconnects AI

@UCMhPRKnK8S_ruoGSt3vm1sQ - 5.7K subscribers

Video essays on AI research and interviews with leading scientists in the field. Linking important ideas of AI. The border between high-level and technical thinking.

Home Videos Live Playlists

Why NVIDIA builds their own open models | Nemotron w/ Bryan Catanzaro

Why NVIDIA builds their own open models | Nemotron w/ Bryan Catanzaro Interconnects AI

1.7K views - 2 weeks ago

"All in on building open models in the U.S."

"All in on building open models in the U.S." Interconnects AI

987 views - 3 weeks ago

8 Graphs Telling Today's Story of Open Models

8 Graphs Telling Today's Story of Open Models Interconnects AI

6K views - 1 month ago

Winners and Losers of Open Models in 2025

Winners and Losers of Open Models in 2025 Interconnects AI

731 views - 2 months ago

How We Built a Leading Reasoning Model (Olmo 3)

How We Built a Leading Reasoning Model (Olmo 3) Interconnects AI

9.7K views - 2 months ago

They Built an AGI Lab in 8 Months

They Built an AGI Lab in 8 Months Interconnects AI

2.9K views - 3 months ago

The State of Open Models

The State of Open Models Interconnects AI

4.1K views - 4 months ago

GPT-5's simultaneous excitement and disappointment

GPT-5's simultaneous excitement and disappointment Interconnects AI

1.5K views - 6 months ago

The ATOM Project: American Truly Open Models

The ATOM Project: American Truly Open Models Interconnects AI

676 views - 6 months ago

Ross Taylor, Ex-Llama reasoning lead, on Chinese open models, scaling RL, & the next 6 months in AI

Ross Taylor, Ex-Llama reasoning lead, on Chinese open models, scaling RL, & the next 6 months in AI Interconnects AI

2.2K views - 6 months ago

The art of training a good (reasoning) language model

The art of training a good (reasoning) language model Interconnects AI

7.4K views - 8 months ago

Self-play for Self-driving and where Scaling Reinforcement Learning is Heading with Eugene Vinitsky

Self-play for Self-driving and where Scaling Reinforcement Learning is Heading with Eugene Vinitsky Interconnects AI

1.8K views - 11 months ago

An Unexpected Reinforcement Learning Renaissance

An Unexpected Reinforcement Learning Renaissance Interconnects AI

14.8K views - 1 year ago

OLMo leads on the secrets of training language models (w Dirk Groeneveld, Kyle Lo, & Luca Soldaini)

OLMo leads on the secrets of training language models (w Dirk Groeneveld, Kyle Lo, & Luca Soldaini) Interconnects AI

1.6K views - 1 year ago

How language model post-training is done today

How language model post-training is done today Interconnects AI

12.9K views - 1 year ago

Quick recap on the state of language model reasoning

Quick recap on the state of language model reasoning Interconnects AI

4.7K views - 1 year ago

Finbarr Timbers on the Future of Reinforcement Learning

Finbarr Timbers on the Future of Reinforcement Learning Interconnects AI

1.6K views - 1 year ago

Tim Dettmers on Open-source AI, LMs, SWE Bench, Agents, Quantization, & Optimization

Tim Dettmers on Open-source AI, LMs, SWE Bench, Agents, Quantization, & Optimization Interconnects AI

1.7K views - 1 year ago

Andrew Carr on Pushing the Boundaries of Generative AI (Beyond Text)

Andrew Carr on Pushing the Boundaries of Generative AI (Beyond Text) Interconnects AI

419 views - 1 year ago

Arvind Narayanan on making sense of AI hype

Arvind Narayanan on making sense of AI hype Interconnects AI

1K views - 1 year ago

Andrew Trask on changing how LLMs store and access information

Andrew Trask on changing how LLMs store and access information Interconnects AI

393 views - 1 year ago

Riley Goodside on OpenAI's o1 and the frontier of prompting LLMs

Riley Goodside on OpenAI's o1 and the frontier of prompting LLMs Interconnects AI

946 views - 1 year ago

Llama 3.2 vision vs. Molmo: Open multimodal models

Llama 3.2 vision vs. Molmo: Open multimodal models Interconnects AI

461 views - 1 year ago

OpenAI's Strawberry and Spending More on Inference

OpenAI's Strawberry and Spending More on Inference Interconnects AI

218 views - 1 year ago

OLMoE and training better foundation models

OLMoE and training better foundation models Interconnects AI

113 views - 1 year ago

Defining open-source AI and open data

Defining open-source AI and open data Interconnects AI

73 views - 1 year ago

Interviewing Ross Taylor on LLM reasoning, Llama post training, and Galactica

Interviewing Ross Taylor on LLM reasoning, Llama post training, and Galactica Interconnects AI

842 views - 1 year ago

Interviewing Sebastian Raschka on the state of open LLMs, Llama 3.1, and AI education

Interviewing Sebastian Raschka on the state of open LLMs, Llama 3.1, and AI education Interconnects AI

803 views - 1 year ago

Llama 3.1 405b and Meta's "open source" AI strategy

Llama 3.1 405b and Meta's "open source" AI strategy Interconnects AI

94 views - 1 year ago

Switching to Claude 3.5 from ChatGPT

Switching to Claude 3.5 from ChatGPT Interconnects AI

714 views - 1 year ago

Interviewing Dean Ball on AI policy: SB-1047, Llama 405B, scaling laws, China, and more

Interviewing Dean Ball on AI policy: SB-1047, Llama 405B, scaling laws, China, and more Interconnects AI

263 views - 1 year ago

Frontiers in LLM synthetic data

Frontiers in LLM synthetic data Interconnects AI

229 views - 1 year ago

How Apple Intelligence Works

How Apple Intelligence Works Interconnects AI

161 views - 1 year ago

A realistic path to robotic foundation models

A realistic path to robotic foundation models Interconnects AI

351 views - 1 year ago

There is no LLM data wall

There is no LLM data wall Interconnects AI

173 views - 1 year ago

OpenAI vs. Scarlett Johansson: Name, image, and AI's likeness

OpenAI vs. Scarlett Johansson: Name, image, and AI's likeness Interconnects AI

177 views - 1 year ago

OpenAI builds Her

OpenAI builds Her Interconnects AI

486 views - 1 year ago

OpenAI's RLHF Specifications

OpenAI's RLHF Specifications Interconnects AI

139 views - 1 year ago

RLHF: A thin line between useful and lobotomized

RLHF: A thin line between useful and lobotomized Interconnects AI

79 views - 1 year ago

Phi 3 and Arctic: Outlier LMs are hints

Phi 3 and Arctic: Outlier LMs are hints Interconnects AI

206 views - 1 year ago

AGI is what you want it to be

AGI is what you want it to be Interconnects AI

132 views - 1 year ago

Llama 3: Scaling open LLMs to AGI

Llama 3: Scaling open LLMs to AGI Interconnects AI

1.3K views - 1 year ago

Stop "reinventing" everything to "solve" AI alignment

Stop "reinventing" everything to "solve" AI alignment Interconnects AI

859 views - 1 year ago

The end of the "best" open LLM

The end of the "best" open LLM Interconnects AI

586 views - 1 year ago

Why we disagree about what open-source AI is

Why we disagree about what open-source AI is Interconnects AI

106 views - 1 year ago

DBRX: The new best open LLM and Databricks' ML strategy

DBRX: The new best open LLM and Databricks' ML strategy Interconnects AI

719 views - 1 year ago

LLM RLHF Evaluations: Trust, performance, and price (bonus, announcing RewardBench)

LLM RLHF Evaluations: Trust, performance, and price (bonus, announcing RewardBench) Interconnects AI

60 views - 1 year ago

LLM commoditization and product moats

LLM commoditization and product moats Interconnects AI

540 views - 1 year ago

Defining an open-source LLM

Defining an open-source LLM Interconnects AI

209 views - 1 year ago

Interviewing Louis Castricato of Synth Labs, Eleuther AI on RLHF, Gemini Drama, DPO, Carper AI

Interviewing Louis Castricato of Synth Labs, Eleuther AI on RLHF, Gemini Drama, DPO, Carper AI Interconnects AI

399 views - 1 year ago

How to cultivate a high-signal AI feed

How to cultivate a high-signal AI feed Interconnects AI

244 views - 1 year ago

Google ships it: Gemma open LLMs and Gemini backlash

Google ships it: Gemma open LLMs and Gemini backlash Interconnects AI

133 views - 1 year ago

10 Sora and Gemini 1.5 follow-ups code-base in context, deepfakes, pixel-peeping, inference costs...

10 Sora and Gemini 1.5 follow-ups code-base in context, deepfakes, pixel-peeping, inference costs... Interconnects AI

134 views - 2 years ago

Releases! OpenAI’s Sora for video, Gemini 1.5's infinite context, and a secret Mistral model

Releases! OpenAI’s Sora for video, Gemini 1.5's infinite context, and a secret Mistral model Interconnects AI

82 views - 2 years ago

Why reward models are still key to understanding LLM alignment

Why reward models are still key to understanding LLM alignment Interconnects AI

401 views - 2 years ago

Alignment-as-a-Service upstarts taking on Scale AI

Alignment-as-a-Service upstarts taking on Scale AI Interconnects AI

98 views - 2 years ago

Open Language Models (OLMos) and the LLM landscape

Open Language Models (OLMos) and the LLM landscape Interconnects AI

218 views - 2 years ago

Model merging lessons in The Waifu Research Department

Model merging lessons in The Waifu Research Department Interconnects AI

182 views - 2 years ago

Local LLMs, some facts some fiction

Local LLMs, some facts some fiction Interconnects AI

68 views - 2 years ago

Multimodal blogging: My AI tools to expand your audience

Multimodal blogging: My AI tools to expand your audience Interconnects AI

281 views - 2 years ago