Karsten Kuhnke PRO
mindchain
AI & ML interests
Industry Grade Humanoid Synthetic Motion Data Generation, Mechanistic Interpretability Data Generation, Sparse Autoencoders, Edge IOT, Gemma Scope 2, RLHF, Edge AI, Alpa SIM, Alpamayo-R1, Cosmos, Isaac SIM, Isaac LAB, GR00T N1.6, Unreal Engine
Recent Activity
reacted
to
their
post
with 🔥
about 17 hours ago
Claude Code Self & Continual Learning
Hey everyone! 👋
30 GitHub Stars in 4 Days - Thank You!
I'm really grateful for the positive response to the Claude Reflect System. In just 4 days, 30 developers have shown interest by starring the project. Thank you so much!
What Is Claude Reflect?
Correct once, never again. Claude Reflect helps Claude Code remember your corrections and preferences across sessions. Instead of repeating the same feedback, the system learns and applies it automatically.
Main Features:
🧠 Learning System
- Detects corrections and preferences from conversations
- Stores them permanently in skill files
- Applies learnings in future sessions
🔒 Safety First
- Automatic backups before changes
- YAML validation
- Git version control
⚡ Two Modes
- Manual: Run /reflect when you want
- Auto: Reflects automatically at session end
How It Works
If you correct Claude to use pytest instead of unittest, this preference gets saved. Next time, Claude will remember and use pytest automatically. It's that simple.
Getting Started
1. Clone the repository
2. Install dependencies
3. Activate the skill
4. Try it out!
The python-project-creator example shows how the system learns from your feedback.
Give It a Try
https://github.com/haddock-development/claude-reflect-system
Feel free to check it out, give feedback, or contribute. Every bit of input helps improve the project!
Thank you so much for your support!
---
#ClaudeCode #AI #MachineLearning #ContinualLearning #OpenSource #Developer #Coding #Python #Productivity #DevTools #GitHub #SoftwareDevelopment #Programming #AIAssistant #DeveloperTools #CodeQuality #Tech
Feel free to give it a try by yourself.
https://github.com/haddock-development/claude-reflect-system
reacted
to
their
post
with 🚀
about 17 hours ago
Claude Code Self & Continual Learning
Hey everyone! 👋
30 GitHub Stars in 4 Days - Thank You!
I'm really grateful for the positive response to the Claude Reflect System. In just 4 days, 30 developers have shown interest by starring the project. Thank you so much!
What Is Claude Reflect?
Correct once, never again. Claude Reflect helps Claude Code remember your corrections and preferences across sessions. Instead of repeating the same feedback, the system learns and applies it automatically.
Main Features:
🧠 Learning System
- Detects corrections and preferences from conversations
- Stores them permanently in skill files
- Applies learnings in future sessions
🔒 Safety First
- Automatic backups before changes
- YAML validation
- Git version control
⚡ Two Modes
- Manual: Run /reflect when you want
- Auto: Reflects automatically at session end
How It Works
If you correct Claude to use pytest instead of unittest, this preference gets saved. Next time, Claude will remember and use pytest automatically. It's that simple.
Getting Started
1. Clone the repository
2. Install dependencies
3. Activate the skill
4. Try it out!
The python-project-creator example shows how the system learns from your feedback.
Give It a Try
https://github.com/haddock-development/claude-reflect-system
Feel free to check it out, give feedback, or contribute. Every bit of input helps improve the project!
Thank you so much for your support!
---
#ClaudeCode #AI #MachineLearning #ContinualLearning #OpenSource #Developer #Coding #Python #Productivity #DevTools #GitHub #SoftwareDevelopment #Programming #AIAssistant #DeveloperTools #CodeQuality #Tech
Feel free to give it a try by yourself.
https://github.com/haddock-development/claude-reflect-system
reacted
to
their
post
with ❤️
about 17 hours ago
Claude Code Self & Continual Learning
Hey everyone! 👋
30 GitHub Stars in 4 Days - Thank You!
I'm really grateful for the positive response to the Claude Reflect System. In just 4 days, 30 developers have shown interest by starring the project. Thank you so much!
What Is Claude Reflect?
Correct once, never again. Claude Reflect helps Claude Code remember your corrections and preferences across sessions. Instead of repeating the same feedback, the system learns and applies it automatically.
Main Features:
🧠 Learning System
- Detects corrections and preferences from conversations
- Stores them permanently in skill files
- Applies learnings in future sessions
🔒 Safety First
- Automatic backups before changes
- YAML validation
- Git version control
⚡ Two Modes
- Manual: Run /reflect when you want
- Auto: Reflects automatically at session end
How It Works
If you correct Claude to use pytest instead of unittest, this preference gets saved. Next time, Claude will remember and use pytest automatically. It's that simple.
Getting Started
1. Clone the repository
2. Install dependencies
3. Activate the skill
4. Try it out!
The python-project-creator example shows how the system learns from your feedback.
Give It a Try
https://github.com/haddock-development/claude-reflect-system
Feel free to check it out, give feedback, or contribute. Every bit of input helps improve the project!
Thank you so much for your support!
---
#ClaudeCode #AI #MachineLearning #ContinualLearning #OpenSource #Developer #Coding #Python #Productivity #DevTools #GitHub #SoftwareDevelopment #Programming #AIAssistant #DeveloperTools #CodeQuality #Tech
Feel free to give it a try by yourself.
https://github.com/haddock-development/claude-reflect-system
Organizations
Nvidia Nemotron Orchestrator
Facebook - Roberta
-
FacebookAI/roberta-base
Fill-Mask • 0.1B • Updated • 8.37M • • 546 -
FacebookAI/xlm-roberta-large-finetuned-conll03-german
Token Classification • Updated • 6.54k • • 14 -
FacebookAI/xlm-roberta-large-finetuned-conll02-spanish
Fill-Mask • 0.6B • Updated • 36 • 2 -
FacebookAI/xlm-roberta-large
Fill-Mask • 0.6B • Updated • 3.4M • • 484
Nvidia Physical AI
Qwen 3 VL - RAG - Reranker
Facebook - Research Plan Dataset
Nvidia Clara - Medical
Nvidia Clara - Molecular
Nvidia Nemotron RAG - Embeddings
-
nvidia/llama-nemotron-embed-vl-1b-v2
Feature Extraction • 2B • Updated • 1.4k • 12 -
nvidia/llama-nemotron-embed-1b-v2
Feature Extraction • 1B • Updated • 16.8k • 29 -
nvidia/llama-embed-nemotron-8b
Feature Extraction • 8B • Updated • 399k • 116 -
nvidia/NV-Embed-v2
Feature Extraction • 8B • Updated • 27.9k • 497
Diffusion Transformer - Video + Audio
-
Lightricks/LTX-2
Image-to-Video • Updated • 736k • • 804 -
Masked Audio Generation using a Single Non-Autoregressive Transformer
Paper • 2401.04577 • Published • 44 -
LTX-2: Efficient Joint Audio-Visual Foundation Model
Paper • 2601.03233 • Published • 94 -
YOLO-World: Real-Time Open-Vocabulary Object Detection
Paper • 2401.17270 • Published • 43
Nvidia Nemotron Cascade
-
nvidia/Nemotron-Cascade-8B
Text Generation • 8B • Updated • 5.49k • 50 -
nvidia/Nemotron-Cascade-8B-Thinking
Text Generation • 8B • Updated • 1.61k • 31 -
nvidia/Nemotron-Cascade-14B-Thinking
Text Generation • 15B • Updated • 3.31k • 55 -
nvidia/Nemotron-Cascade-8B-Intermediate-ckpts
Text Generation • Updated • 10
Google Gemma 3 - LiteRT Models
-
google/functiongemma-270m-it
Text Generation • 0.3B • Updated • 81.2k • 786 -
litert-community/embeddinggemma-300m
Sentence Similarity • Updated • 1.63k • 31 -
google/gemma-3n-E2B-it-litert-lm
Text Generation • Updated • 19.1k • 266 -
google/gemma-3n-E4B-it-litert-lm
Text Generation • Updated • 20.1k • 277
Nvidia Cosmos 2 - Cosmos-Transfer 2.5
Robotics
Nvidia Nemotron RL Datasets - NeMo Gym
-
nvidia/Nemotron-RL-knowledge-web_search-mcqa
Viewer • Updated • 2.93k • 356 • 6 -
nvidia/Nemotron-RL-agent-workplace_assistant
Viewer • Updated • 1.8k • 300 • 9 -
nvidia/Nemotron-RL-instruction_following
Preview • Updated • 154 • 7 -
nvidia/Nemotron-RL-instruction_following-structured_outputs
Viewer • Updated • 9.95k • 308 • 25
Nvidia Nemotron RAG Datasets
Gemma 3N - Mobile multimordal Edition
Small Thinking Models
Self-Correcting Delta Transformer
Self-Correcting Delta Transformer - DDL provides the Hardware mechanism (The Erazor), NL solves the software problem.
META - SAM
Edge LLMs - Tiny but STRONG
Nvidia Nemotron - Persona Datasets
Nvidia Nemotron - Reward Models
LLM as a judge
-
nvidia/Llama-3.3-Nemotron-70B-Reward-Principle
Text Generation • 71B • Updated • 92 • 6 -
nvidia/Qwen3-Nemotron-32B-GenRM-Principle
Text Generation • 33B • Updated • 699 • 11 -
nvidia/Qwen3-Nemotron-32B-RLBFF
Text Generation • 33B • Updated • 65 • 27 -
nvidia/Qwen3-Nemotron-8B-BRRM
Text Generation • Updated • 149 • 8
Embeddings
Translation at the Edge
Qwen - Long Reasoning
OCR
-
PaddlePaddle/PaddleOCR-VL
Image-Text-to-Text • 1.0B • Updated • 12k • 1.47k -
baidu/ERNIE-4.5-0.3B-Paddle
Text Generation • 0.4B • Updated • 88 • 19 -
baidu/ERNIE-4.5-21B-A3B-Paddle
Text Generation • 22B • Updated • 68 • 13 -
ibm-granite/granite-docling-258M
Image-Text-to-Text • 0.3B • Updated • 212k • 1.08k
Iquenst Loopcoder
Automatic Speech Recognition
Mobile App Engine Models
Hybrid Attention Models
Datasets Pretraining - Nemotron V3
Mamba/Transformers Combo Hybride
Byte Level Models
Video Analysis
Diffusion Language Models
Video Generation
Graphics
VL-JEPA
Google Gemma Scope 2 - Neuronpedia
Google Gemma Scope 2: JumpReLU SAEs for Gemma 3 interpretability. 270M PT/IT, 1B PT variants. Neuronpedia integration. Mechanistic analysis.
Google Gemma - Quantized
Quantized Gemma 3 models: QAT for efficient deployment. Gemma-3-27B-IT Q4. Low memory, fast inference. Edge & production-ready LLMs.
Reward Models
NVIDIA Nemotron reward models: 340B, 8B BRRM, 70B/32B principle-based. RLHF training, preference learning, AI alignment research.
T5 Gemma 2
T5Gemma-2 encoder-decoder models: 270M, 1B, 4B sizes. Text-to-text, summarization, translation. Google's architecture for structured generation.
-
google/t5gemma-2-270m-270m
Image-Text-to-Text • 0.8B • Updated • 18k • 163 -
google/t5gemma-2-4b-4b
Image-Text-to-Text • 9B • Updated • 10.1k • 135 -
google/t5gemma-2-1b-1b
Image-Text-to-Text • 2B • Updated • 13.3k • 64 -
T5Gemma 2: Seeing, Reading, and Understanding Longer
Paper • 2512.14856 • Published • 1
Mamba/Transformers Combo Hybride
Hybrid Mamba + Transformer architectures. NVIDIA Nemotron-3-Nano-30B-A3B (32B). BF16 & GGUF. Efficient long-context & in-context learning.
-
nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16
Text Generation • 32B • Updated • 320k • 554 -
bartowski/nvidia_Nemotron-3-Nano-30B-A3B-GGUF
Text Generation • 32B • Updated • 10.4k • 9 -
nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-Base-BF16
Text Generation • 32B • Updated • 12.1k • 89 -
nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-FP8
Text Generation • 32B • Updated • 725k • • 232
Best Model to fit into dual RTX 6000 Build
Hugging Face LeRobot - Pi0Fast
Google - Embedding Gemma
Nvidia Thor + Rasberry + Oak 4D Dual Build
Qwen 3 VL - RAG - Embeddings
Nvidia Nemotron - Content Safety
Nvidia Clara - Biology
Nvidia Nemotron RAG - Reranking
Nvidia - Alpamayo-R1
-
nvidia/Alpamayo-R1-10B
Robotics • 11B • Updated • 13.8k • 263 -
nvidia/PhysicalAI-Autonomous-Vehicles
Updated • 152k • 667 -
nvidia/PhysicalAI-Autonomous-Vehicles-NuRec
Updated • 7.37k • 80 -
Alpamayo-R1: Bridging Reasoning and Action Prediction for Generalizable Autonomous Driving in the Long Tail
Paper • 2511.00088 • Published • 2
Nvidia Nemotron Speech
-
nvidia/nemotron-speech-streaming-en-0.6b
Automatic Speech Recognition • Updated • 2.69k • 303 -
nvidia/parakeet-tdt-0.6b-v3
Automatic Speech Recognition • Updated • 73.6k • 534 -
nvidia/parakeet_realtime_eou_120m-v1
Updated • 746 • 106 -
nvidia/multitalker-parakeet-streaming-0.6b-v1
Automatic Speech Recognition • Updated • 632 • 61
GPT OSS + Steering
Nvidia Cosmos 2 - Cosmos-Reason 2
NVIDIA Cosmos 2 - Cosmos-Predict 2.5
Smartphone
Nvidia Nemotron Savety Datasets
-
nvidia/Nemotron-AIQ-Agentic-Safety-Dataset-1.0
Viewer • Updated • 10.8k • 1.37k • 10 -
nvidia/Nemotron-Content-Safety-Reasoning-Dataset
Preview • Updated • 68 • 4 -
nvidia/Aegis-AI-Content-Safety-Dataset-2.0
Viewer • Updated • 33.4k • 3.03k • 71 -
nvidia/Nemotron-Content-Safety-Audio-Dataset
Viewer • Updated • 1.93k • 1.15k • 3
Nvidia Nemotron VLM Datasets
Deep Thinking Models
Small Coders
YOLO
Cheep Coding Plan APIs
RLM - Neuro-Symbolic Architecture - Reasonig Traces
Inference Wrapper - Models: Root LLM (The Architect) + Python REPL (The Engine) + Sub LLMs (The Workers) Spaned by querys
Nvidia Nemotron - Post Training Datasets
Nvidia Nemotron - Pre Traininng Datasets
Topological Transformer - Deepseek
: Manifold-Constrained Hyper-Connections
Deep Research
PP-StructureV3
Circuit Sparsity
Text to Motion
TTS Models
Audio Segmenting - Meta SAM 3 Audio
Datasets Post Training Nemotron V3
Mamba/Transformers Combo Hybride
All OPEN Models
Image to 3D
Deep Research Specialized Models
IBM Granite
Small OCR
Hierarchical Reinforcement Learning
Bread&Butter
Top LLMs 2025: ZAI GLM-4.7 (358B) & Moonshot Kimi-K2-Thinking. Next-gen reasoning, code, multilingual. State-of-the-art performance. Production-ready.
Haddock Custom Sparse Autodecoders
Custom JumpReLU Sparse Autoencoders for mechanistic interpretability. T5Gemma-2 SAEs across all layers. AI safety & interpretability research.
Nemo-Gym
NVIDIA Nemotron RL datasets for AI agent training. Web search, workplace tasks, instruction following, structured outputs. RLHF & alignment research.
-
nvidia/Nemotron-RL-knowledge-web_search-mcqa
Viewer • Updated • 2.93k • 356 • 6 -
nvidia/Nemotron-RL-agent-workplace_assistant
Viewer • Updated • 1.8k • 300 • 9 -
nvidia/Nemotron-RL-instruction_following
Preview • Updated • 154 • 7 -
nvidia/Nemotron-RL-instruction_following-structured_outputs
Viewer • Updated • 9.95k • 308 • 25
Trained
Custom-trained models by mindchain: reward models & SAEs. Haddock Reward Mini (8B), function-specific SAEs. AI interpretability & alignment research.
Auto Decoders
JumpReLU SAEs for Gemma 3 interpretability. EleutherAI models, DeepSeek-R1, Pythia SAEs. Mechanistic interpretability & AI safety research.
FunctionGemma (Gemma 3)
Function calling: Google FunctionGemma-270m-IT & Mobile Actions dataset (9.65k). Efficient tool use in small LMs. AI agent development.
Hyper Graph Reasoning
Best Model to fit into dual RTX 6000 Build
Nvidia Nemotron Orchestrator
Hugging Face LeRobot - Pi0Fast
Facebook - Roberta
-
FacebookAI/roberta-base
Fill-Mask • 0.1B • Updated • 8.37M • • 546 -
FacebookAI/xlm-roberta-large-finetuned-conll03-german
Token Classification • Updated • 6.54k • • 14 -
FacebookAI/xlm-roberta-large-finetuned-conll02-spanish
Fill-Mask • 0.6B • Updated • 36 • 2 -
FacebookAI/xlm-roberta-large
Fill-Mask • 0.6B • Updated • 3.4M • • 484
Google - Embedding Gemma
Nvidia Physical AI
Nvidia Thor + Rasberry + Oak 4D Dual Build
Qwen 3 VL - RAG - Reranker
Qwen 3 VL - RAG - Embeddings
Facebook - Research Plan Dataset
Nvidia Nemotron - Content Safety
Nvidia Clara - Medical
Nvidia Clara - Biology
Nvidia Clara - Molecular
Nvidia Nemotron RAG - Reranking
Nvidia Nemotron RAG - Embeddings
-
nvidia/llama-nemotron-embed-vl-1b-v2
Feature Extraction • 2B • Updated • 1.4k • 12 -
nvidia/llama-nemotron-embed-1b-v2
Feature Extraction • 1B • Updated • 16.8k • 29 -
nvidia/llama-embed-nemotron-8b
Feature Extraction • 8B • Updated • 399k • 116 -
nvidia/NV-Embed-v2
Feature Extraction • 8B • Updated • 27.9k • 497
Nvidia - Alpamayo-R1
-
nvidia/Alpamayo-R1-10B
Robotics • 11B • Updated • 13.8k • 263 -
nvidia/PhysicalAI-Autonomous-Vehicles
Updated • 152k • 667 -
nvidia/PhysicalAI-Autonomous-Vehicles-NuRec
Updated • 7.37k • 80 -
Alpamayo-R1: Bridging Reasoning and Action Prediction for Generalizable Autonomous Driving in the Long Tail
Paper • 2511.00088 • Published • 2
Diffusion Transformer - Video + Audio
-
Lightricks/LTX-2
Image-to-Video • Updated • 736k • • 804 -
Masked Audio Generation using a Single Non-Autoregressive Transformer
Paper • 2401.04577 • Published • 44 -
LTX-2: Efficient Joint Audio-Visual Foundation Model
Paper • 2601.03233 • Published • 94 -
YOLO-World: Real-Time Open-Vocabulary Object Detection
Paper • 2401.17270 • Published • 43
Nvidia Nemotron Speech
-
nvidia/nemotron-speech-streaming-en-0.6b
Automatic Speech Recognition • Updated • 2.69k • 303 -
nvidia/parakeet-tdt-0.6b-v3
Automatic Speech Recognition • Updated • 73.6k • 534 -
nvidia/parakeet_realtime_eou_120m-v1
Updated • 746 • 106 -
nvidia/multitalker-parakeet-streaming-0.6b-v1
Automatic Speech Recognition • Updated • 632 • 61
Nvidia Nemotron Cascade
-
nvidia/Nemotron-Cascade-8B
Text Generation • 8B • Updated • 5.49k • 50 -
nvidia/Nemotron-Cascade-8B-Thinking
Text Generation • 8B • Updated • 1.61k • 31 -
nvidia/Nemotron-Cascade-14B-Thinking
Text Generation • 15B • Updated • 3.31k • 55 -
nvidia/Nemotron-Cascade-8B-Intermediate-ckpts
Text Generation • Updated • 10
GPT OSS + Steering
Google Gemma 3 - LiteRT Models
-
google/functiongemma-270m-it
Text Generation • 0.3B • Updated • 81.2k • 786 -
litert-community/embeddinggemma-300m
Sentence Similarity • Updated • 1.63k • 31 -
google/gemma-3n-E2B-it-litert-lm
Text Generation • Updated • 19.1k • 266 -
google/gemma-3n-E4B-it-litert-lm
Text Generation • Updated • 20.1k • 277
Nvidia Cosmos 2 - Cosmos-Reason 2
Nvidia Cosmos 2 - Cosmos-Transfer 2.5
NVIDIA Cosmos 2 - Cosmos-Predict 2.5
Robotics
Smartphone
Nvidia Nemotron RL Datasets - NeMo Gym
-
nvidia/Nemotron-RL-knowledge-web_search-mcqa
Viewer • Updated • 2.93k • 356 • 6 -
nvidia/Nemotron-RL-agent-workplace_assistant
Viewer • Updated • 1.8k • 300 • 9 -
nvidia/Nemotron-RL-instruction_following
Preview • Updated • 154 • 7 -
nvidia/Nemotron-RL-instruction_following-structured_outputs
Viewer • Updated • 9.95k • 308 • 25
Nvidia Nemotron Savety Datasets
-
nvidia/Nemotron-AIQ-Agentic-Safety-Dataset-1.0
Viewer • Updated • 10.8k • 1.37k • 10 -
nvidia/Nemotron-Content-Safety-Reasoning-Dataset
Preview • Updated • 68 • 4 -
nvidia/Aegis-AI-Content-Safety-Dataset-2.0
Viewer • Updated • 33.4k • 3.03k • 71 -
nvidia/Nemotron-Content-Safety-Audio-Dataset
Viewer • Updated • 1.93k • 1.15k • 3
Nvidia Nemotron RAG Datasets
Nvidia Nemotron VLM Datasets
Gemma 3N - Mobile multimordal Edition
Deep Thinking Models
Small Thinking Models
Small Coders
Self-Correcting Delta Transformer
Self-Correcting Delta Transformer - DDL provides the Hardware mechanism (The Erazor), NL solves the software problem.
YOLO
META - SAM
Cheep Coding Plan APIs
Edge LLMs - Tiny but STRONG
RLM - Neuro-Symbolic Architecture - Reasonig Traces
Inference Wrapper - Models: Root LLM (The Architect) + Python REPL (The Engine) + Sub LLMs (The Workers) Spaned by querys
Nvidia Nemotron - Persona Datasets
Nvidia Nemotron - Post Training Datasets
Nvidia Nemotron - Reward Models
LLM as a judge
-
nvidia/Llama-3.3-Nemotron-70B-Reward-Principle
Text Generation • 71B • Updated • 92 • 6 -
nvidia/Qwen3-Nemotron-32B-GenRM-Principle
Text Generation • 33B • Updated • 699 • 11 -
nvidia/Qwen3-Nemotron-32B-RLBFF
Text Generation • 33B • Updated • 65 • 27 -
nvidia/Qwen3-Nemotron-8B-BRRM
Text Generation • Updated • 149 • 8
Nvidia Nemotron - Pre Traininng Datasets
Embeddings
Topological Transformer - Deepseek
: Manifold-Constrained Hyper-Connections
Translation at the Edge
Deep Research
Qwen - Long Reasoning
PP-StructureV3
OCR
-
PaddlePaddle/PaddleOCR-VL
Image-Text-to-Text • 1.0B • Updated • 12k • 1.47k -
baidu/ERNIE-4.5-0.3B-Paddle
Text Generation • 0.4B • Updated • 88 • 19 -
baidu/ERNIE-4.5-21B-A3B-Paddle
Text Generation • 22B • Updated • 68 • 13 -
ibm-granite/granite-docling-258M
Image-Text-to-Text • 0.3B • Updated • 212k • 1.08k
Circuit Sparsity
Iquenst Loopcoder
Text to Motion
Automatic Speech Recognition
TTS Models
Mobile App Engine Models
Audio Segmenting - Meta SAM 3 Audio
Hybrid Attention Models
Datasets Post Training Nemotron V3
Mamba/Transformers Combo Hybride
Datasets Pretraining - Nemotron V3
Mamba/Transformers Combo Hybride
All OPEN Models
Byte Level Models
Image to 3D
Video Analysis
Deep Research Specialized Models
Diffusion Language Models
IBM Granite
Video Generation
Small OCR
Graphics
Hierarchical Reinforcement Learning
VL-JEPA
Bread&Butter
Top LLMs 2025: ZAI GLM-4.7 (358B) & Moonshot Kimi-K2-Thinking. Next-gen reasoning, code, multilingual. State-of-the-art performance. Production-ready.
Google Gemma Scope 2 - Neuronpedia
Google Gemma Scope 2: JumpReLU SAEs for Gemma 3 interpretability. 270M PT/IT, 1B PT variants. Neuronpedia integration. Mechanistic analysis.
Haddock Custom Sparse Autodecoders
Custom JumpReLU Sparse Autoencoders for mechanistic interpretability. T5Gemma-2 SAEs across all layers. AI safety & interpretability research.
Google Gemma - Quantized
Quantized Gemma 3 models: QAT for efficient deployment. Gemma-3-27B-IT Q4. Low memory, fast inference. Edge & production-ready LLMs.
Nemo-Gym
NVIDIA Nemotron RL datasets for AI agent training. Web search, workplace tasks, instruction following, structured outputs. RLHF & alignment research.
-
nvidia/Nemotron-RL-knowledge-web_search-mcqa
Viewer • Updated • 2.93k • 356 • 6 -
nvidia/Nemotron-RL-agent-workplace_assistant
Viewer • Updated • 1.8k • 300 • 9 -
nvidia/Nemotron-RL-instruction_following
Preview • Updated • 154 • 7 -
nvidia/Nemotron-RL-instruction_following-structured_outputs
Viewer • Updated • 9.95k • 308 • 25
Reward Models
NVIDIA Nemotron reward models: 340B, 8B BRRM, 70B/32B principle-based. RLHF training, preference learning, AI alignment research.
Trained
Custom-trained models by mindchain: reward models & SAEs. Haddock Reward Mini (8B), function-specific SAEs. AI interpretability & alignment research.
T5 Gemma 2
T5Gemma-2 encoder-decoder models: 270M, 1B, 4B sizes. Text-to-text, summarization, translation. Google's architecture for structured generation.
-
google/t5gemma-2-270m-270m
Image-Text-to-Text • 0.8B • Updated • 18k • 163 -
google/t5gemma-2-4b-4b
Image-Text-to-Text • 9B • Updated • 10.1k • 135 -
google/t5gemma-2-1b-1b
Image-Text-to-Text • 2B • Updated • 13.3k • 64 -
T5Gemma 2: Seeing, Reading, and Understanding Longer
Paper • 2512.14856 • Published • 1
Auto Decoders
JumpReLU SAEs for Gemma 3 interpretability. EleutherAI models, DeepSeek-R1, Pythia SAEs. Mechanistic interpretability & AI safety research.
Mamba/Transformers Combo Hybride
Hybrid Mamba + Transformer architectures. NVIDIA Nemotron-3-Nano-30B-A3B (32B). BF16 & GGUF. Efficient long-context & in-context learning.
-
nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16
Text Generation • 32B • Updated • 320k • 554 -
bartowski/nvidia_Nemotron-3-Nano-30B-A3B-GGUF
Text Generation • 32B • Updated • 10.4k • 9 -
nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-Base-BF16
Text Generation • 32B • Updated • 12.1k • 89 -
nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-FP8
Text Generation • 32B • Updated • 725k • • 232
FunctionGemma (Gemma 3)
Function calling: Google FunctionGemma-270m-IT & Mobile Actions dataset (9.65k). Efficient tool use in small LMs. AI agent development.