Runtime error
š
Sam Joshua
SamJoshua
AI & ML interests
None yet
Recent Activity
reacted
to
garrethlee's
post
with š„
15 days ago
The latest o1 model from OpenAI is still unable to answer 9.11 > 9.9 correctly š¤
A possible explanation? Tokenization - and our latest work investigates how it affects a model's ability to do math!
In this blog post, we discuss:
š¢ The different ways numbers are tokenized in modern LLMs
š§Ŗ Our detailed approach in comparing these various methods
š„Ŗ How we got a free boost in arithmetic performance by adding a few lines of code to the base Llama 3 tokenizer
š and a definitive, best tokenization method for math in LLMs!
Check out our work here: https://huggingface.co/spaces/huggingface/number-tokenization-blog
Organizations
None yet
spaces
1
models
9
SamJoshua/bert-finetuned-sem_eval-english
Text Classification
ā¢
Updated
ā¢
10
SamJoshua/EsperBERTo
Fill-Mask
ā¢
Updated
ā¢
5
SamJoshua/phi-1_5-finetuned-gsm8k
Text Generation
ā¢
Updated
ā¢
17
SamJoshua/llama-7b-dolly
Text Generation
ā¢
Updated
ā¢
11
SamJoshua/llama2-qlora-orca
Updated
ā¢
2
SamJoshua/llama2-qlora-french
Updated
ā¢
2
SamJoshua/gpt-neo-125M
Updated
SamJoshua/ppo-Huggy
Reinforcement Learning
ā¢
Updated
ā¢
11
SamJoshua/MoonLanding
Reinforcement Learning
ā¢
Updated
datasets
None public yet