Kunvar Thaman

Taking apart neural networks and putting them back together for a living.

prev @si_pbc and @Akamai

[Uzay @uzpg_]

There is a relatively neglected opportunity to understand and train models to have better behavior, potentially optimizing less for frontier capability. This “good behavior” encapsulates many traits - truthfulness, the ability to understand/adapt to a user in various ways, not being overly reward-seeking, etc.

[Kunvar Thaman @kunvar_]

@sambhavgupta159 How are all the annotation companies charging $40/hr still in business lol

[calix @calixo888]

Argus: an open-source robotics data annotation and quality pipeline.

With new frontier VLMs like GPT-6 Astra, we can generate rich, high-quality annotations for robotics data for both training and dataset analysis. Argus delivers detailed, timestamp-level annotations while catching issues like mislabeled instructions, sped-up recordings, swapped camera streams, and unflagged operator mistakes.

To support better data quality for the robotics community, we’re open-sourcing Argus for anyone to use 🧵

[Kunvar Thaman @kunvar_]

Argus increases the effective value of a training sample 20-30x.

We're moving from robotics as a field being data constrained to one which is compute constrained

[calix @calixo888]

Argus: an open-source robotics data annotation and quality pipeline.

With new frontier VLMs like GPT-6 Astra, we can generate rich, high-quality annotations for robotics data for both training and dataset analysis. Argus delivers detailed, timestamp-level annotations while catching issues like mislabeled instructions, sped-up recordings, swapped camera streams, and unflagged operator mistakes.

To support better data quality for the robotics community, we’re open-sourcing Argus for anyone to use 🧵

[calix @calixo888]

Read the full analysis at https://t.co/keW7O5WFaL, and run Argus on your data at https://t.co/MW3rQcDt9F. For collaborations or questions, contact us at [email protected].

Credit to @ericli_ for undertaking this project.

[calix @calixo888]

Argus is a model-agnostic pipeline and can run on any sufficiently capable VLM. In our test comparisons, Astra and GPT-6.1 Sol produced the densest annotations, which we’ve found is highly correlated with overall accuracy.

[Kunvar Thaman @kunvar_]

this is so cool!

[Fulcrum @fulcrum_inc]

Echo is trained to embody the style of whatever persona you choose. You can try it for free at https://t.co/M7epfJLArx.

[Kunvar Thaman @kunvar_]

Good writing is hard, the Fulcrum team is very smart - check out Echo!

[Fulcrum @fulcrum_inc]

Say hello to Echo, the best writing model at style imitation. Echo beats frontier models at writing tasks ranging from fiction to technical explanations, despite costing less than $5K to train.

[Fulcrum @fulcrum_inc]

Say hello to Echo, the best writing model at style imitation. Echo beats frontier models at writing tasks ranging from fiction to technical explanations, despite costing less than $5K to train.

[finbarr @finbarrtimbers]

Great to see more open data quality work. Data quality work doesn't get nearly enough attention in the open source/weights community, but it's almost all of the work in industry.

[Kunvar Thaman @kunvar_]

@kushal1t Everyone knows you should look at the data, no one wants to do it though. Lots of alpha in just looking

[Kunvar Thaman @kunvar_]

The unreasonable effectiveness of actually looking at data

[calix @calixo888]

We (@PantheonInc) have been working on a robotics data quality pipeline that uncovered a series of major problems in public robotics datasets, especially for world modeling.

To improve the quality of data available to open-source robotics, we're publishing annotations for four of the most popular datasets. Some examples of issues, and our report 🧵

[calix @calixo888]

We (@PantheonInc) have been working on a robotics data quality pipeline that uncovered a series of major problems in public robotics datasets, especially for world modeling.

To improve the quality of data available to open-source robotics, we're publishing annotations for four of the most popular datasets. Some examples of issues, and our report 🧵

[Kenneth Ge @kennyge0]

Forget skill files. What if your model could switch its own finetuning adapters for each task?
I gave Qwen 35B a tool to switch its own adapter in a multi-part task. It beat out subagents, Arrow, and normal fine-tuning, used up to 46x fewer tokens, and incurred ~0 capability tax.

[Mark Khrapko @markkhrapko]

a 10 minute application for a $50,000 fellowship. no strings.

a total of $1,000,000 going out, towards making more magic in the world.

presenting the Basis Fellowship.

run by @markkhrapko ,@thewildstevenp , and spencer green

[Kunvar Thaman @kunvar_]

would recommend applying!

[Mark Khrapko @markkhrapko]

a 10 minute application for a $50,000 fellowship. no strings.

a total of $1,000,000 going out, towards making more magic in the world.

presenting the Basis Fellowship.

run by @markkhrapko ,@thewildstevenp , and spencer green

[Kunvar Thaman @kunvar_]

@bonniesjli @GoogleDeepMind Nice, excited :)

[Kunvar Thaman @kunvar_]

@vamshibon Haven't decided on one yet - there's a bunch on meetup but happy to plan/organize one

[Kunvar Thaman @kunvar_]

or go on runs and hikes!

[Kunvar Thaman @kunvar_]

I’ll be in Seoul for ICML from July 6–12. If anyone’s around and wants to meet up, grab coffee, or go for a walk, hit me up!