finetune-smolLM2-360M-Instruct
This model is a fine-tuned version of HuggingFaceTB/SmolLM2-360M-Instruct on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 2.7585
Model description
SmolLM2 is a family of compact language models available in three size: 135M, 360M, and 1.7B parameters. They are capable of solving a wide range of tasks while being lightweight enough to run on-device.
SmolLM2 demonstrates significant advances over its predecessor SmolLM1, particularly in instruction following, knowledge, reasoning. The 360M model was trained on 4 trillion tokens using a diverse dataset combination: FineWeb-Edu, DCLM, The Stack, along with new filtered datasets we curated and will release soon. We developed the instruct version through supervised fine-tuning (SFT) using a combination of public datasets and our own curated datasets. We then applied Direct Preference Optimization (DPO) using UltraFeedback.
The instruct model additionally supports tasks such as text rewriting, summarization and function calling thanks to datasets developed by Argilla such as Synth-APIGen-v0.1.
Intended uses & limitations
SmolLM2 models primarily understand and generate content in English. They can produce text on a variety of topics, but the generated content may not always be factually accurate, logically consistent, or free from biases present in the training data. These models should be used as assistive tools rather than definitive sources of information. Users should always verify important information and critically evaluate any generated content.
Training and evaluation data
https://huggingface.co/datasets/Salesforce/wikitext The WikiText language modeling dataset is a collection of over 100 million tokens extracted from the set of verified Good and Featured articles on Wikipedia. The dataset is available under the Creative Commons Attribution-ShareAlike License.
Compared to the preprocessed version of Penn Treebank (PTB), WikiText-2 is over 2 times larger and WikiText-103 is over 110 times larger. The WikiText dataset also features a far larger vocabulary and retains the original case, punctuation and numbers - all of which are removed in PTB. As it is composed of full articles, the dataset is well suited for models that can take advantage of long term dependencies.
Each subset comes in two different variants:
Raw (for character level work) contain the raw tokens, before the addition of the (unknown) tokens. Non-raw (for word level work) contain only the tokens in their vocabulary (wiki.train.tokens, wiki.valid.tokens, and wiki.test.tokens). The out-of-vocabulary tokens have been replaced with the the token.
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 1
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 2.7017 | 0.0871 | 50 | 2.8763 |
| 2.6474 | 0.1743 | 100 | 2.8244 |
| 2.5094 | 0.2614 | 150 | 2.8044 |
| 2.6284 | 0.3486 | 200 | 2.7901 |
| 2.7183 | 0.4357 | 250 | 2.7798 |
| 2.6457 | 0.5229 | 300 | 2.7732 |
| 2.7641 | 0.6100 | 350 | 2.7691 |
| 2.6276 | 0.6972 | 400 | 2.7661 |
| 2.7211 | 0.7843 | 450 | 2.7639 |
| 2.6556 | 0.8715 | 500 | 2.7603 |
| 2.7031 | 0.9586 | 550 | 2.7587 |
Framework versions
- Transformers 4.45.1
- Pytorch 2.4.0
- Datasets 3.0.1
- Tokenizers 0.20.0
Citation
@misc{allal2024SmolLM2, title={SmolLM2 - with great data, comes great performance}, author={Loubna Ben Allal and Anton Lozhkov and Elie Bakouch and Gabriel Martín Blázquez and Lewis Tunstall and Agustín Piqueres and Andres Marafioti and Cyril Zakka and Leandro von Werra and Thomas Wolf}, year={2024}, }