Katerina Brujeva PRO

KrazyKitty

1 14

james47kjv's profile picture

Johnblick187's profile picture

webbrain-7c94e1b2's profile picture

https://krazykitty.dev

krazykitty2503

AI & ML interests

Independent AI & language-model research • Transformer architectures • Tokenisation & embeddings • Dataset engineering & provenance • Reproducible training pipelines • Model evaluation • Inference optimisation • AI agents • Responsible AI • Python & PyTorch • KittyLM development.

Recent Activity

liked a Space about 3 hours ago

KrazyKitty/KittyLM-Research

updated a Space 4 days ago

kitty-tweak-labs/ktl-research-hub

published a Space 4 days ago

kitty-tweak-labs/ktl-research-hub

View all activity

Organizations

KrazyKitty 's collections 4

NEKO//AGENT — AGENT SYSTEMS

Reasoning with actions, learned tool use, reflection, agent-computer interfaces, and executable agent evaluation.

Sleeping

Agents

NEKO//AGENT Research

🐾

Evidence-led research for safe, tool-using agents
ReAct: Synergizing Reasoning and Acting in Language Models

Paper • 2210.03629 • Published Oct 6, 2022 • 38
Toolformer: Language Models Can Teach Themselves to Use Tools

Paper • 2302.04761 • Published Feb 9, 2023 • 12
Reflexion: Language Agents with Verbal Reinforcement Learning

Paper • 2303.11366 • Published Mar 20, 2023 • 9

KRAZYKITTY // MODEL TRAINING

Compute-aware pretraining, parameter-efficient adaptation, quantised finetuning, and preference optimisation.

Running

1

KittyLM Research

🐱

1

Modular language-model research from first principles
Training Compute-Optimal Large Language Models

Paper • 2203.15556 • Published Mar 29, 2022 • 13
LoRA: Low-Rank Adaptation of Large Language Models

Paper • 2106.09685 • Published Jun 17, 2021 • 67
QLoRA: Efficient Finetuning of Quantized LLMs

Paper • 2305.14314 • Published May 23, 2023 • 64

KRAZYKITTY // DATA PROVENANCE & EVALUATION

Transparent corpus construction, source governance, filtering research, benchmark documentation, and contamination-aware evaluation.

KrazyKitty/Data-Lab

Updated 4 days ago • 46
Dolma: an Open Corpus of Three Trillion Tokens for Language Model Pretraining Research

Paper • 2402.00159 • Published Jan 31, 2024 • 66
The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scale

Paper • 2406.17557 • Published Jun 25, 2024 • 108
DataComp-LM: In search of the next generation of training sets for language models

Paper • 2406.11794 • Published Jun 17, 2024 • 56

KRAZYKITTY // TRANSFORMER FOUNDATIONS

Architecture and modelling foundations spanning attention, bidirectional representation, autoregressive scaling, and compact open baselines.

Running

1

KittyLM Research

🐱

1

Modular language-model research from first principles
Attention Is All You Need

Paper • 1706.03762 • Published Jun 12, 2017 • 146
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

Paper • 1810.04805 • Published Oct 11, 2018 • 33
Language Models are Few-Shot Learners

Paper • 2005.14165 • Published May 28, 2020 • 21

NEKO//AGENT — AGENT SYSTEMS

Reasoning with actions, learned tool use, reflection, agent-computer interfaces, and executable agent evaluation.

Sleeping

Agents

NEKO//AGENT Research

🐾

Evidence-led research for safe, tool-using agents
ReAct: Synergizing Reasoning and Acting in Language Models

Paper • 2210.03629 • Published Oct 6, 2022 • 38
Toolformer: Language Models Can Teach Themselves to Use Tools

Paper • 2302.04761 • Published Feb 9, 2023 • 12
Reflexion: Language Agents with Verbal Reinforcement Learning

Paper • 2303.11366 • Published Mar 20, 2023 • 9

KRAZYKITTY // DATA PROVENANCE & EVALUATION

Transparent corpus construction, source governance, filtering research, benchmark documentation, and contamination-aware evaluation.

KrazyKitty/Data-Lab

Updated 4 days ago • 46
Dolma: an Open Corpus of Three Trillion Tokens for Language Model Pretraining Research

Paper • 2402.00159 • Published Jan 31, 2024 • 66
The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scale

Paper • 2406.17557 • Published Jun 25, 2024 • 108
DataComp-LM: In search of the next generation of training sets for language models

Paper • 2406.11794 • Published Jun 17, 2024 • 56

KRAZYKITTY // MODEL TRAINING

Compute-aware pretraining, parameter-efficient adaptation, quantised finetuning, and preference optimisation.

Running

1

KittyLM Research

🐱

1

Modular language-model research from first principles
Training Compute-Optimal Large Language Models

Paper • 2203.15556 • Published Mar 29, 2022 • 13
LoRA: Low-Rank Adaptation of Large Language Models

Paper • 2106.09685 • Published Jun 17, 2021 • 67
QLoRA: Efficient Finetuning of Quantized LLMs

Paper • 2305.14314 • Published May 23, 2023 • 64

KRAZYKITTY // TRANSFORMER FOUNDATIONS

Architecture and modelling foundations spanning attention, bidirectional representation, autoregressive scaling, and compact open baselines.

Running

1

KittyLM Research

🐱

1

Modular language-model research from first principles
Attention Is All You Need

Paper • 1706.03762 • Published Jun 12, 2017 • 146
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

Paper • 1810.04805 • Published Oct 11, 2018 • 33
Language Models are Few-Shot Learners

Paper • 2005.14165 • Published May 28, 2020 • 21

Katerina Brujeva PRO

AI & ML interests

Recent Activity

Organizations

KrazyKitty 's collections 4

NEKO//AGENT Research

KittyLM Research

KittyLM Research

NEKO//AGENT Research

KittyLM Research

KittyLM Research