SmolLM2-360M Intermediate Checkpoints
We are releasing an intermediate checkpoint of SmolLM2 to enable further research on mechanistic interpretability and learning dynamics. This repo contains the checkpoint every 160000 steps which correspond to ~250B tokens.
How to Load a Checkpoint
# pip install transformers
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
checkpoint = "HuggingFaceTB/SmolLM2-360M-intermediate-checkpoints"
revision = "step-160000" # replace by the revision you want
device = torch.device("cuda" if torch.cuda.is_available() else "mps" if hasattr(torch, 'mps') and torch.mps.is_available() else "cpu")
tokenizer = AutoTokenizer.from_pretrained(checkpoint, revision=revision)
model = AutoModelForCausalLM.from_pretrained(checkpoint, revision=revision).to(device)
inputs = tokenizer.encode("Gravity is", return_tensors="pt").to(device)
outputs = model.generate(inputs)
print(tokenizer.decode(outputs[0]))
Training Details
For comprehensive information about SmolLM2 training methodology, please refer to:
- Our model page
- Our GitHub repository
- Our paper
Checkpoint Details
The GBS (Global Batch Size) in token of SmolLM2-360M is 1 572 864. To get the number of tokens knowing the step you need to do:
nb_tokens = nb_step * GBS
License
Citation
@misc{allal2025smollm2smolgoesbig,
title={SmolLM2: When Smol Goes Big -- Data-Centric Training of a Small Language Model},
author={Loubna Ben Allal and Anton Lozhkov and Elie Bakouch and Gabriel Martín Blázquez and Guilherme Penedo and Lewis Tunstall and Andrés Marafioti and Hynek Kydlíček and Agustín Piqueres Lajarín and Vaibhav Srivastav and Joshua Lochner and Caleb Fahlgren and Xuan-Son Nguyen and Clémentine Fourrier and Ben Burtenshaw and Hugo Larcher and Haojun Zhao and Cyril Zakka and Mathieu Morlon and Colin Raffel and Leandro von Werra and Thomas Wolf},
year={2025},
eprint={2502.02737},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2502.02737},
}