Downloads · 30 days
26
9% of all-time downloads
toksuite/meta-llama-Llama-3.2-1B-seed_777_model_seed_222
meta-llama-Llama-3.2-1B-seed_777_model_seed_222 is a text generation model from toksuite. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as mit.
<p align="left" <img src="./toksuite-logo.png" alt="TokSuite Logo" width="260"/ </p
Downloads · 30 days
26
9% of all-time downloads
All-time downloads
303
Public
Parameters
1.8B
3.6 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors3.6 GB · 100%
From the Hugging Face model README
TokSuite–Llama-3.2 is part of TokSuite, a suite of language models designed to study the impact of tokenizer choice on language model behavior under controlled conditions.
This model uses the Llama-3.2 tokenizer and is otherwise identical to the other TokSuite-Llama models
(toksuite/meta-llama-Llama-3.2-1B and
toksuite/meta-llama-Llama-3.2-1B-seed_777_model_seed_888)
in architecture, training data, training budget and only differ in the initialization, denoted by the model_seed.
Processing details:
Llama-3.2 was included in TokSuite to represent a multilingual BPE tokenizer with modern GPT-4–style pretokenization and a medium-to-large vocabulary size. As described in the tokenizer selection rationale of the TokSuite paper, Llama-3.2 exemplifies a contemporary tokenizer design that combines subword segmentation with preprocessing conventions used in recent large language models.
Including Llama-3.2 enables TokSuite to study tokenizer behavior in settings where:
This makes Llama-3.2 a representative example of modern multilingual BPE tokenization.
The architecture and training setup are identical across all TokSuite models; only the tokenizer differs.
The model was trained on a multilingual corpus totaling approximately 100B tokens, composed of:
You can find the pretraining dataset here: toksuite/toksuite_pretraining_data
All TokSuite models are trained using a fixed token budget, following common practice in large-scale language model training.
TokSuite–Llama-3.2 variants are evaluated on the TokSuite robustness benchmark, which measures sensitivity to real-world text perturbations, including:
Tokenization Robustness under Multilingual Text Perturbations
Values represent relative performance drop, computed as (Acc_clean − Acc_perturbed) / Acc_clean, where lower values indicate greater robustness.
Perturbation types include:
NEN denotes non-English inputs and EN denotes English inputs. The Avg column reports the average relative performance drop across all perturbation categories.
| Model | Input (NEN) | Diacr. (NEN) | Orth. & Gram. (EN) | Orth. & Gram. (NEN) | Morph (EN) | Morph (NEN) | Noise (EN) | Noise (NEN) | LaTeX (EN) | STEM (EN) | Unic. (EN) | Avg ↓ |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| toksuite/meta-llama-Llama-3.2-1B-seed_777_model_seed_888 | 0.31 | 0.44 | 0.12 | 0.15 | 0.25 | 0.13 | 0.09 | 0.24 | 0.09 | 0.26 | 0.58 | 0.24 |
| toksuite/meta-llama-Llama-3.2-1B | 0.31 | 0.56 | 0.11 | 0.11 | 0.25 | 0.10 | 0.09 | 0.26 | 0.13 | 0.29 | 0.60 | 0.26 |
| toksuite/meta-llama-Llama-3.2-1B-seed_777_model_seed_222 | 0.29 | 0.54 | 0.09 | 0.12 | 0.20 | 0.13 | 0.14 | 0.29 | 0.19 | 0.37 | 0.59 | 0.27 |
| Avg | 0.30 | 0.51 | 0.11 | 0.12 | 0.24 | 0.12 | 0.10 | 0.26 | 0.13 | 0.31 | 0.59 | 0.25 |
This model is intended for:
It is not instruction-tuned, aligned, or optimized for deployment.
TokSuite models are released to support scientific investigation of tokenization effects.
They may reflect biases present in large-scale web data and should not be used in high-stakes or user-facing applications without additional safeguards.
If you use this model, please cite:
@article{toksuite2025,
title={TokSuite: Measuring the Impact of Tokenizer Choice on Language Model Behavior},
author={Altıntaş, Gul Sena and Ehghaghi, Malikeh and Lester, Brian and Liu, Fengyuan and Zhao, Wanru and Ciccone, Marco and Raffel, Colin},
year={2025},
arxiv={https://arxiv.org/abs/2512.20757},
}