tysam-code / hlb-gptLinks
Minimalistic, extremely fast, and hackable researcher's toolbench for GPT models in 307 lines of code. Reaches <3.8 validation loss on wikitext-103 on a single A100 in <100 seconds. Scales to larger models with one parameter change (feature currently in alpha).
☆355Updated last year
Alternatives and similar repositories for hlb-gpt
Users that are interested in hlb-gpt are comparing it to the libraries listed below
Sorting:
- Fast bare-bones BPE for modern tokenizer training☆175Updated 7 months ago
- ☆94Updated 2 years ago
- Language Modeling with the H3 State Space Model☆522Updated 2 years ago
- a small code base for training large models☆322Updated 9 months ago
- Ungreedy subword tokenizer and vocabulary trainer for Python, Go & Javascript☆616Updated last year
- A crude RLHF layer on top of nanoGPT with Gumbel-Softmax trick☆294Updated 2 years ago
- The repository for the code of the UltraFastBERT paper☆519Updated last year
- ☆144Updated 2 years ago
- JAX implementation of the Llama 2 model☆216Updated 2 years ago
- ☆316Updated last year
- git extension for {collaborative, communal, continual} model development☆217Updated last year
- ☆416Updated 2 years ago
- A puzzle to learn about prompting☆135Updated 2 years ago
- Helpers and such for working with Lambda Cloud☆52Updated 2 years ago
- Puzzles for exploring transformers☆384Updated 2 years ago
- Simple Transformer in Jax☆142Updated last year
- Extract full next-token probabilities via language model APIs☆248Updated last year
- Fully fine-tune large models like Mistral, Llama-2-13B, or Qwen-14B completely for free☆232Updated last year
- MiniHF is an inference, human preference data collection, and fine-tuning tool for local language models. It is intended to help the user…☆183Updated 3 months ago
- An interactive exploration of Transformer programming.☆271Updated 2 years ago
- Fine-tune mistral-7B on 3090s, a100s, h100s☆725Updated 2 years ago
- Comprehensive analysis of difference in performance of QLora, Lora, and Full Finetunes.☆83Updated 2 years ago
- batched loras☆349Updated 2 years ago
- Minimal example scripts of the Hugging Face Trainer, focused on staying under 150 lines☆196Updated last year
- A repository for log-time feedforward networks☆224Updated last year
- ☆289Updated last year
- Open weights language model from Google DeepMind, based on Griffin.☆661Updated 2 weeks ago
- Full finetuning of large language models without large memory requirements☆94Updated 4 months ago
- A comprehensive deep dive into the world of tokens☆226Updated last year
- Inference code for Persimmon-8B☆412Updated 2 years ago