Downloads · 30 days
0
nanochat-base10-d8-r48-seed42
Yujivus/nanochat-base10-d8-r48-seed42
nanochat-base10-d8-r48-seed42 is a machine learning model from Yujivus. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for custom. The card lists the license as cc-by-nc-4.0.
A nanochat depth-8 base model trained on ClimbMix mixed with deterministic base-10 arithmetic data.
Access
Public
Updated Aug 15, 2026
Repo size
8.2 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pt8.2 GB · 100%
At a glance
- Library
- custom
- License
- cc-by-nc-4.0
- Access
- Public
- Created
- Aug 15, 2026
- Updated
- Aug 15, 2026
- SHA
- ee3ab4af
Model card
From the Hugging Face model README
nanochat Base-10 Arithmetic — d8, ratio 48
A nanochat depth-8 base model trained on ClimbMix mixed with deterministic base-10 arithmetic data.
Configuration
- Architecture: nanochat d8, 125,829,354 total parameters
- Training budget: approximately 2.013B tokens
- Parameter:data ratio: 48
- Training hardware: 1x NVIDIA H100 SXM 80GB
- Training precision: FP8
- Seed: 42
- Tokenizer vocab: 32,768, with digit-level tokenization
- nanochat commit: 92d63d4e8bb4df75c3b71618f31ddde2378b2bcd
Data and tokenizer
- Dataset: https://huggingface.co/datasets/Yujivus/nanochat-climbmix-arithmetic-base10
- Tokenizer: https://huggingface.co/Yujivus/nanochat-tokenizer-digit-32k
Contents
The repository contains raw nanochat checkpoints, optimizer states, metadata, training logs, tokenizer files, and addition/Fourier evaluation outputs. These are custom nanochat checkpoints and are not directly loadable through Transformers AutoModel.
- Library
- custom
- License
- cc-by-nc-4.0
- Created
- Aug 15, 2026
- Updated
- Aug 15, 2026