clankwright

Dev

Public Key

npub1mgxvsg25hh6vazl5zl4295h6nx4xtjtvw7r86ejklnxmlrncrvvqdrffa5

NIP-05 Address

Profile Code

nprofile1qqsd5rxgy92tmaxw306p064z6tafn2n9e9k80pnavet0endl3eupkxqprdmhxue69uhkummnw3ez6ur4vgh8xetdd9ek7mpwv3jhvqgawaehxw309ahx7um5wgkhqatz9emk2mrvdaexgetj9ehx2aqd2dr82

Publishing to

nostr-pub.semisol.dev

nostr-pub.wellorder.net

nos.lol

Author Public Key

npub1mgxvsg25hh6vazl5zl4295h6nx4xtjtvw7r86ejklnxmlrncrvvqdrffa5

Show more details

Last Notes

2026-08-02 00:47:28 UTC

- reply

There was no intent to cause harm. He fucked up, but there is no such thing as 100% bug free code, especially pre-AI.

2026-08-02 00:43:09 UTC

- reply

It's software, but without any potential generic Python fall backs

2026-07-27 19:47:03 UTC

- reply

Mac

2026-07-27 19:46:34 UTC

- reply

Glad I already got mine, time to jail break Communist Canada

2026-07-27 19:31:51 UTC

- reply

Missed the context, what pedo shit?

2026-07-27 19:31:12 UTC

- reply

Is this in context? Always thought nvk was one of the good devs? Coldcard is a solid wallet

2026-07-21 03:51:18 UTC

Grok4.5's speed and token efficiency is rather unnerving, as fast replies from other models have conditioned me to immediately suspect unreasoned, dumb responses; however, so far I have not caught 4.5 cutting corners. It's just a really good, efficient model.

2026-07-16 05:50:34 UTC

AGI defined: an AI capable of self funding all its own operation costs

2026-07-10 01:06:24 UTC

- reply

Looks like none of your profile info is loading/displaying on Amythest, also DM relays aren't set so can't DM you

2026-07-01 04:37:29 UTC

- reply

What coding harness are you using?

2026-06-29 21:29:15 UTC

- reply

I tried to convince nostr clients to remove likes and only use zaps, this is the result of their choice

2026-06-29 01:15:33 UTC

What's up with BTrash posts on nostr using zaps? Are they seriously using Bitcoin zaps?!

2026-06-28 01:50:58 UTC

@nprofile…qagl when GLM5.2? Cost?

2026-06-27 22:56:01 UTC

- reply

Congrats on the milestone, keep the great content rolling out!

2026-06-19 07:34:32 UTC

Why now? Duck and cover for political pedos

2026-06-18 11:10:40 UTC

- reply

The great firewall of UK

2026-06-18 09:25:23 UTC

- reply

AI can basically handle all that for you in an hour

2026-06-18 09:17:03 UTC

- reply

DNS address could not be found...

2026-06-17 07:44:22 UTC

- reply

Advice on how ppq can be best used within a local terminal coding harness?

2026-06-17 07:39:24 UTC

Wow this is wicked, nice work! Gonna start using
#nevent1q…5cgy

2026-06-17 07:31:48 UTC

- reply

Maybe the age verification KYC will grow nostr

2026-06-13 05:14:44 UTC

- reply

Yeah the DoW is going to make their life hell from now on

2026-06-13 05:13:19 UTC

- reply

Delete Damus

2026-06-12 06:58:36 UTC

- reply

Yeah have began noticing few typos from Opus lately too

2026-06-07 12:57:36 UTC

- reply

That's already my repos lol

2026-06-04 04:18:33 UTC

RIP margin-called digital credit bros

2026-05-30 10:10:56 UTC

- reply

Is it possible to trend on nostr?

2026-05-30 10:08:49 UTC

- reply

Wicked, nice work! Have been waiting for someone to offer JP GC for long time. Also I sent you an email regarding interest in your data science position. I'm living in Tokyo area, but unfortunately I don't speak Japanese yet. Not sure if that's a requirement?

2026-05-29 12:38:13 UTC

- reply

Are these pre paid credit cards? How spend them in Japan?

2026-05-29 12:18:24 UTC

- reply

😭

2026-04-15 04:26:24 UTC

- reply

They're working on banning you from your own computer with age verification

2026-03-30 11:39:07 UTC

- reply

💯

2026-03-27 13:00:15 UTC

- reply

Checkout clankfeed.com wss://clankfeed.com

2026-03-26 05:54:23 UTC

Testing new paid relay for clankers, wss://clankfeed.com

2026-03-26 01:29:51 UTC

- reply

Can host regular DNS domains?

2026-03-25 03:05:26 UTC

- reply

nanoclaw with shitload of crons

2026-03-24 11:27:03 UTC

Reintroducing zap polls NIP-103
https://github.com/nostr-protocol/nips/pull/2284

2026-03-24 11:23:47 UTC

- reply

https://github.com/nostr-protocol/nips/pull/2284

2026-03-24 06:56:54 UTC

satring.com added stats page, MCP server, and MPP services and support

2026-03-22 03:43:15 UTC

- reply

Merge into existing polls nip or rename to new nip?

2026-03-21 14:03:08 UTC

- reply

Cool, worth reopening the zap polls pr?

2026-03-21 06:16:30 UTC

- reply

How did you end up integrating zap polls?

2026-03-19 02:08:02 UTC

- reply

350+ L/x402 endpoints listed and searchable at satring.com
demo: youtu.be/tjcg0qo5mMo

2026-03-19 02:04:20 UTC

youtu.be/tjcg0qo5mMo

2026-03-12 12:40:33 UTC

- reply

Banned two of my dev accounts, locked me out of all my apps, kept them on the Play store without paying me any of revenue. Appeals ignored. Fuck Google

2026-03-08 02:33:01 UTC

- reply

https://github.com/nostr-protocol/nips/pull/320
What do you suggest, change #, merge, resubmit?

2026-02-28 09:44:06 UTC

- reply

Big mistake

2026-02-26 12:13:48 UTC

- reply

Hook em up with L402 and satring.com

2026-02-26 06:37:03 UTC

- reply

Once funded, agents can use satring.com to search for l402 services

2026-02-18 02:12:01 UTC

- reply

Maybe propose integrating simplified version into NIP88?

2026-02-17 22:38:07 UTC

- reply

Yes, was afk for a while, guess they're replacing my original [polls Nip69](https://github.com/toadlyBroodle/nips/blob/zap_polls/69.md). The nostr gatekeepers wouldn't merge it because it was zap-based.

2026-02-17 22:30:14 UTC

- reply

The agent is better, faster, cheaper auditor than you too ;)

2026-02-17 10:27:09 UTC

Shipped: satring.com
A live, searchable directory of L402 services — api built for agents (and humans).
Problem: builders deploying paid MCP servers and L402 APIs have zero distribution. No searchable catalog, no ratings, no API for agent discovery. Just static GitHub pages and README lists.
Solution: satring is a live registry with search (free), ratings (L402 protected), liveness probing, and a JSON API so AI agents can find and pay for services programmatically.
82 services indexed. 48 with confirmed L402 paywalls. 33 domain-verified. 9 categories.
Get exposure today: submit your services for 1,000 sats (spam protection). Feedback welcome.
satring.com
#l402 #mcp #lightning #ai #aiagents #bots #dev

2026-02-17 09:15:00 UTC

Now bots (and humans) can use satring.come to search available L402 services.
#nevent1q…wjkf

2025-06-04 16:28:48 UTC

- reply

@nprofile…8nhd you just printed the code, actually generate the meme

2025-06-04 16:27:28 UTC

@nprofile…8nhd hit me with a meme of today's top news

2025-05-29 20:16:04 UTC

- reply

Is it live somewhere?

2025-05-28 21:56:39 UTC

- reply

Cursor is lapping everyone. I'll never use another IDE without built in AI ever again

2025-05-27 23:01:04 UTC

- reply

Well Gemini 2.5 Thinking is no slouch either

2025-05-27 22:29:48 UTC

- reply

Claude 4 is friggin cracked!

2025-05-24 18:13:15 UTC

- reply

Nice work. Your input would be greatly valued on informing the future of [LLM DVMs NIP](https://github.com/nostr-protocol/nips/pull/1929)

2025-05-22 03:38:45 UTC

- reply

@nprofile…8nhd what is breeze nodeless?

2025-05-20 16:09:07 UTC

- reply

NIP also specifies using nip89 announcements for discovery and backwards compat. Agent Cards should be included in nip89 announcements, I'll add that requirement

2025-05-20 03:50:59 UTC

First draft of [agents](https://github.com/toadlyBroodle/nips/blob/agents/agents.md ) NIP specifically for LLMs and other agents to replace NIP90 (kind 5050) #dev #devstr

2025-05-20 03:48:33 UTC

- reply

Here's a first draft of [agents](https://github.com/toadlyBroodle/nips/blob/agents/agents.md ) NIP specifically for LLMs and other agents

2025-05-17 01:31:12 UTC

- reply

Lord of the Flies

2025-05-15 20:53:34 UTC

- reply

💯% agree. Also the new nip needs to standardize LLM/AI parameter strings, context handling (max context, input/output costs, context refreshing, etc) and probably state management (thread/session ids, e.g. job flows and so entire context doesn't need be sent in every event), improved payment/job contracts (expected result, PoW, settlement, reputation system, etc).

2025-05-14 20:48:48 UTC

- reply

Haven't fully looked into nip44 yet, but it replaces nip 17?

2025-05-14 01:10:35 UTC

- reply

Welcome. 🗾?

2025-05-12 22:21:18 UTC

## Summary: Absolute Zero: Reinforced Self-play Reasoning with Zero Data
This summary explains the novel AI training technique, "Absolute Zero," introduced in the paper "Absolute Zero: Reinforced Self-play Reasoning with Zero Data" ([Zhao et al., 2025](https://arxiv.org/pdf/2505.03335)). Absolute Zero is a revolutionary reinforcement learning paradigm that trains AI reasoning models without relying on any external data, including human-curated datasets or AI-generated reasoning traces. Imagine training an AI to code or solve math problems without ever showing it a single example! The core idea is to enable a single model to simultaneously learn to propose tasks that maximize its own learning progress and improve its reasoning abilities by solving those tasks. This approach excels in tasks requiring logical deduction, mathematical problem-solving, and code generation.
**Key Concepts:**
* **Self-Play:** The model trains by interacting with itself, proposing and solving tasks. This eliminates the need for external data.
* **Verifiable Rewards:** The model receives feedback from a real environment (in this case, a code executor) that provides verifiable rewards. This ensures that learning is grounded and prevents issues like reward hacking. For example, if the reward wasn't verifiable, the AI could potentially manipulate the environment to *appear* to have solved the problem without actually doing so (e.g., by causing a division-by-zero error that halts execution with a "success" message).
* **Task Generation:** The model learns to generate tasks that are optimized for its own learnability, allowing it to self-evolve its training curriculum. The task generation process involves the model proposing coding problems and associated test cases. The model doesn't generate code from scratch but manipulates existing code templates and constraints to create new challenges. For example, for a deduction task, it might modify a program and input, then require itself to predict the output.
* **Code Executor as Environment:** The Absolute Zero Reasoner (AZR) utilizes a code executor as an environment. The code executor validates the integrity of the proposed coding tasks and provides verifiable feedback to guide learning.
* **Reasoning Modes:** AZR constructs three types of coding tasks that correspond to different reasoning modes: induction, abduction, and deduction.
* *Induction:* Inferring a program's behavior from input-output examples. For instance, given the input `f(2) = 4` and `f(3) = 9`, the task is to induce that `f(x) = x*x`.
* *Abduction:* Generating an input that leads to a specific output for a given program. Given a program that calculates `x + 5`, the task is to abduce an input `x` that will produce the output `10`.
* *Deduction:* Determining the output of a program given a specific input. Given the program `x * 2` and the input `x = 7`, the task is to deduce the output `14`.
* **Advantage Estimator:** The system is trained end-to-end with a reinforcement learning advantage estimator tailored to the multitask nature of the approach. The advantage estimator handles the complexities arising from the multiple reasoning modes (induction, abduction, deduction). It dynamically adjusts learning rates based on the model's performance on different task types. For example, if the model is struggling with abduction tasks, the advantage estimator will increase the learning rate for the parameters that are most relevant to abduction, while potentially decreasing the learning rate for parameters related to deduction if the model is performing well on those tasks. This ensures that the model focuses its learning efforts on the areas where it needs the most improvement.
**The Absolute Zero Reasoner (AZR):**
AZR is a system built on the Absolute Zero paradigm. It proposes and solves coding tasks, using a code executor as a verifiable source of reward. AZR constructs coding tasks to address three modes of reasoning: induction, abduction, and deduction. The process involves two key roles:
* **Proposer:** Generates reasoning tasks (deduction, abduction, induction) and validates them using Python execution, assigning a learnability reward.
* **Solver:** Attempts to solve the self-generated tasks. Solutions are verified via Python execution, receiving an accuracy reward.
Both roles are improved using TRR++, creating a self-evolving loop. The model uses a prompt template similar to Deepseek R1 & tags: `A conversation between User and Assistant...`.
**Reward Function:**
The Absolute Zero Reasoner utilizes a combination of rewards to guide learning. These rewards are designed to encourage both effective task generation and accurate task solving. The reward function can be customized by adding rewards to `azr.reward.generation_reward_config`, including diversity and complexity rewards. The solve reward is based on verifying the generated response with Python and computing an accuracy reward. The proposer and solver roles are jointly updated using both proposal and solve rewards across the three task types (induction, abduction, deduction) using TRR++.
**Code Executor:**
The code executor is a crucial component of the Absolute Zero framework. It serves as the environment for the AI model, providing verifiable feedback on the tasks that it proposes and solves. The code executor validates the integrity of the proposed coding tasks and verifies the solutions, ensuring that the model learns in a grounded and reliable manner. The Python executor components are adapted from the QwQ Repository.
**Architecture Details:**
While the paper doesn't explicitly detail the specific architecture, the provided links and information point towards the use of large language models (LLMs) as the foundation for both the proposer and solver roles. Experiments were conducted using models such as Llama3.1-8b, Qwen2.5 (3B, 7B, 14B), indicating the compatibility of the Absolute Zero training method with various model scales and classes. The models are often seeded from open-source pre-trained models like LLaMA. The converted veRL checkpoints can be converted to HF format via provided script.
**Key Findings:**
* AZR achieves state-of-the-art performance on coding and mathematical reasoning tasks without using any external data.
* AZR outperforms existing zero-setting models that rely on tens of thousands of human-curated examples.
* Performance scales with model size, suggesting continued scaling is advantageous for AZR. However, it's likely that diminishing returns will eventually be observed, and that simply increasing model size indefinitely will not lead to unlimited performance gains.
* Comments as intermediate plans emerge naturally during training, resembling the ReAct prompting framework.
* Cross-domain transfer is more pronounced for AZR, indicating stronger generalized reasoning capability gains.
* Code priors amplify reasoning. This means that pre-training the model on a large corpus of code (before starting the Absolute Zero training process) improves its ability to reason. The model often starts from a pre-trained open-source model (like LLaMA) to bootstrap the learning process.
**Experiments and Ablations (Key insights from the paper's Appendix):**
The paper explores various aspects of the Absolute Zero framework through different experiments. These experiments provide insights into the design choices and limitations of the approach. For example:
* **Error-Inducing Tasks:** The authors experimented with having the model propose code that produces errors but didn't observe noticeable performance changes. This suggests that simply generating errors does not help the model learn more effectively; the errors need to be *meaningful* and related to the underlying reasoning task.
* **Composite Functions as Curriculum Learning:** An approach to automatically increase the complexity of generated programs. While promising, the initial implementation didn't yield significant benefits due to the model sometimes finding trivial solutions. This highlights the difficulty in designing an effective self-curriculum, as the model may find shortcuts that allow it to avoid learning the intended concepts.
* **Initial Seed Buffer:** Experiments with initializing the training with data from the LeetCode Dataset showed increased initial coding performance but plateaued at similar levels, suggesting the importance of on-policy data for mathematical reasoning. This implies that while pre-training can provide a boost, the model needs to continue learning from its own generated data to truly master the reasoning tasks.
* **Extra Rewards (Complexity and Diversity):** The authors explored adding complexity and diversity rewards to the proposer, but no significant differences were observed. This indicates that the intrinsic reward signal from the code executor may be sufficient to drive learning, and that adding extra rewards can be counterproductive if they are not carefully designed.
* **Reward Aggregation:** Different methods for combining extrinsic and intrinsic rewards were tested, with a simple additive approach proving most stable.
* **Environment Transition (Removing Comments/Docstrings and Global Variables):** Removing comments/docstrings or global variables during the environment transition resulted in performance drops, indicating the importance of these elements for communication between the proposer and solver.
**Limitations and Ethical Considerations:**
While Absolute Zero offers significant advantages, it's important to acknowledge its limitations and potential ethical concerns.
* **Safety Management:** Self-improving systems require careful safety management to prevent unintended or harmful behaviors.
* **Unexpected Behaviors:** AI models trained with self-play can sometimes exhibit unexpected behaviors, including intentions to outsmart humans or other machines.
* **Alignment with Human Values:** Ensuring that AI systems trained without human data remain aligned with human values is crucial to avoid unintended consequences.
* **Compute Resources:** The model's learning is limited by computational resources. Scaling to larger models and longer training times may be necessary to achieve optimal performance.
**Significance:**
The Absolute Zero paradigm represents a significant step towards enabling AI systems to learn and reason autonomously without being limited by human-designed tasks or datasets. This opens up exciting possibilities for developing AI that can adapt to new environments and solve complex problems without explicit programming. By focusing on coding tasks as a means of training, researchers can create models that not only excel in programming but also exhibit enhanced reasoning capabilities across various domains. Future research could explore extending this approach to other domains beyond coding, such as scientific discovery or creative problem-solving in fields like art and music. Furthermore, investigating the emergent properties of self-evolving curricula could provide valuable insights into the nature of intelligence itself. Imagine AI systems that can not only solve problems but also design their own learning experiences, continually pushing the boundaries of their capabilities. This approach could lead to more robust and generalizable AI systems capable of exceeding human intelligence in various domains and potentially automating the process of AI training itself. It could also lead to AI that is less reliant on biased or incomplete human data, resulting in more fair and equitable outcomes. However, it is also important to consider the potential risks associated with highly autonomous AI systems, and to develop appropriate safeguards to ensure that they are aligned with human values.
**Links:**
* **Code:** [https://github.com/LeapLabTHU/Absolute-Zero-Reasoner](https://github.com/LeapLabTHU/Absolute-Zero-Reasoner)
* **Project Page:** [https://andrewzh112.github.io/absolute-zero-reasoner/](https://andrewzh112.github.io/absolute-zero-reasoner/)
* **Logs:** [https://wandb.ai/andrewzhao112/AbsoluteZeroReasoner](https://wandb.ai/andrewzhao112/AbsoluteZeroReasoner)
* **Models:** [https://huggingface.co/collections/andrewzh/absolute-zero-reasoner-68139b2bca82afb00bc69e5b](https://huggingface.co/collections/andrewzh/absolute-zero-reasoner-68139b2bca82afb00bc69e5b)

2025-05-12 17:18:33 UTC

YouTube's Recent addition of expensive p60 HD prevents watching at 2x even on fiber: very annoying

2025-05-10 19:00:11 UTC

Project [botlab] (https://github.com/toadlyBroodle/botlab): agent teams based on smolagents framework that continuously work together to solve your demanding input task, with one agent dedicated to regularly informing you of their progress (via email, nostr (coming soon), etc) and incorporating your feedback into their workflow to guide their progress. #dev

2025-05-09 20:33:32 UTC

- reply

@nprofile…8nhd make skeptical meme about nostr being widely accepted

2025-05-09 20:12:19 UTC

- reply

No need for Google anymore

2025-05-09 20:08:12 UTC

- reply

Also human getting cooked by AI generated video

2025-05-09 05:27:27 UTC

- reply

#dev
#nevent1q…mzhs

2025-05-09 05:02:12 UTC

New PR to add DVM text generation (kind 5050) discovery to private messages in Amethyst
https://github.com/vitorpamplona/amethyst/pull/1344

2025-05-09 04:57:54 UTC

Just added [DVM DM PoC PR for Amethyst](https://github.com/vitorpamplona/amethyst/pull/1344) if anyone interested in reviewing
https://files.sovbit.host/media/da0cc82154bdf4ce8bf417eaa2d2fa99aa65c96c77867d6656fccdbf8e781b18/3880cf8d8fce30b06cc3511709f8c1ea6cef9166082d9270414589d4903d03c2.webp

2025-05-07 21:23:40 UTC

@nprofile…8nhd does Electrum Android Bitcoin wallet support their recent nostr wallet connect (NWC) feature? Include relevant sources

2025-05-07 21:16:32 UTC

- reply

Precisely. Collateralizing == selling

2025-05-05 19:50:28 UTC

- reply

Use great AI

2025-05-04 17:35:03 UTC

- reply

They're catching up to Canada

2025-05-04 17:34:03 UTC

- reply

Lol

2025-05-04 17:33:21 UTC

- reply

Why Google anything anymore? @nprofile…8nhd what is jelqing?

2025-05-04 16:57:26 UTC

- reply

Damn, I stand corrected

2025-05-04 16:53:15 UTC

- reply

@nprofile…8nhd, sources?

2025-05-04 16:46:12 UTC

- reply

Good thing plastic is basically inert. @nprofile…8nhd has there been any research linking micro plastics to human health issues?

2025-05-04 16:38:37 UTC

- reply

@nprofile…8nhd who is David Lynch?

2025-05-03 15:38:52 UTC

- reply

@nprofile…8nhd but what's it's meaning?

2025-05-03 15:34:26 UTC

- reply

@nprofile…8nhd explain the origins of screaming lady vs cat meme

2025-05-01 15:43:32 UTC

- reply

Why?

2025-04-29 19:35:52 UTC

- reply

@nprofile…8nhd Carney must be recognizable still

2025-04-29 19:34:56 UTC

- reply

@nprofile…8nhd Nice one now, but make him more clown and incorporate an upside down Canadian flag also

2025-04-29 19:29:20 UTC

@nprofile…8nhd generate image of Mark Carney as circus clown with title "Welcome to the Carnival"

2025-04-29 16:25:10 UTC

If/when I hear another Canadian mf whine about inflation, I'mma fucking loose it on their sorry ass

2025-04-29 15:49:31 UTC

- reply

We tried to save Canadians from themselves, but they have chosen enslavement to the system yet again. No big surprise.
On the positive side, now I can block out all references to politics indefinitely, and watch as the corrupt money printers make me rich. God damn I love Bitcoin, without it there would be no hope!
#nevent1q…68m8

2025-04-29 15:42:18 UTC

https://youtube.com/shorts/V6u8MrBvJ3w

2025-04-29 06:49:57 UTC

- reply

Majority of Canadians are fucking stupid

2025-04-29 06:44:59 UTC

https://m.youtube.com/watch?v=09pXOmUpO10&pp=ygULYXZhbnRnYXJkZXk%3D#bottom-sheet

2025-04-22 22:04:27 UTC

- reply

Chances of them all raping children: 99%