TroyDoesAI @TroyDoesAI

Open For Work huggingface.co/TroyDoesAI San Mateo Ca Joined April 2024

Tweets

125
Followers

52
Following

55
Likes

86

Eric Hartford @QuixiAI

5 days ago

Wow - Qwen3-Coder-30b AWQ (4bit) on a single 3090, 115 tokens per second. It just zero-shat Pac-Man. It's no GLM4.5-Air - but, it runs on a single 3090!

21 40 391 29K 177

Download Image

Glados + 1.2B Tool Calling that knows how to inference a larger secondary model (BlackSheep 8B) haha Learning tool calling and tuning a model to handle parallel and sequential tool calling is pretty cool for a local Alexa that you can add whatever functions you want it to have.

1 0 1 24 1

Download Video

TroyDoesAI @TroyDoesAI

2 months ago

What if I told you Sycophancy is a Skill Diff not an inherit problem with LLM's?

0 0 1 41 0

Download Gif

TroyDoesAI @TroyDoesAI

3 months ago

0 0 1 50 0

Download Image

TroyDoesAI @TroyDoesAI

3 months ago

The futures gonna be weird. 👨‍🔬

0 0 0 49 0

Download Video

Sakana AI @SakanaAILabs

3 months ago

Introducing Reinforcement-Learned Teachers (RLTs): Transforming how we teach LLMs to reason with reinforcement learning (RL). Blog: sakana.ai/rlt Paper: arxiv.org/abs/2506.08388 Traditional RL focuses on “learning to solve” challenging problems with expensive LLMs and…

25 245 1K 165K 773

Download Video

TroyDoesAI @TroyDoesAI

5 months ago

These new @Alibaba_Qwen models are great! Loving Qwen3 when you unlock them through abliteration. >:P

0 0 1 68 0

Download Gif

Jo Kristian Bergum @jobergum

5 months ago

Billions worth of system prompts are exposed in a single repo. The scale of IP leaks is remarkable in the LLM era.

241 1K 18K 2.3M 25K

Download Image

Karan Dalal @karansdalal

6 months ago

Today, we're releasing a new paper – One-Minute Video Generation with Test-Time Training. We add TTT layers to a pre-trained Transformer and fine-tune it to generate one-minute Tom and Jerry cartoons with strong temporal consistency. Every video below is produced directly by…

187 932 5K 1.4M 3K

Download Video

Eric Hartford @QuixiAI

6 months ago

That's a shocker

Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 @elder_plinius

6 months ago

That's a shocker

174 73 3K 278K 244

Download Image

5 1 26 3K 1

Maxime Labonne @maximelabonne

7 months ago

In the original technique, a refusal direction is computed by comparing the residual streams between target (harmful) and baseline (harmless) samples. Here, the model was abliterated by computing a refusal direction based on hidden states (inspired by Sumandora's repo) for each…

2 3 77 5K 11

Download Image

TroyDoesAI @TroyDoesAI

10 months ago

He has more people on his side than the world knows, (not everyone is vocal) and copycats and vigilante mentality is very strong right now. IDoesAI, seen many startups with shot products, ai is so inaccurate that I can hardly believed a use case was insurance classification. 🖕

0 0 0 44 0

Download Image

TroyDoesAI @TroyDoesAI

11 months ago

BlackSheep-Coder is coming and its hilarious! Note: Found a new method to prevent catastrophic forgetting with using Ablation on models without making it stupid, you can also use this method for continuous training without losing previous skills. 😈

0 0 1 55 0

Download Gif

TroyDoesAI @TroyDoesAI

12 months ago

imgur.com/UlXTAd5

0 0 0 30 0

TroyDoesAI @TroyDoesAI

12 months ago

youtube.com/shorts/QFNV7NL… AI Alignment is Fun 👨‍💻

0 0 0 30 0

TroyDoesAI @TroyDoesAI

a year ago

reddit.com/r/LocalLLaMA/c… If you are looking for something uniquely abliterated/uncensored/unbiased/politically unwoke/morally incorrect, and ethically confused toxic model. HF: TroyDoesAI/BlackSheep-Llama3.2-3B I got the model for you, try out my recent BlackSheep on HuggingFace.