Downloads · 30 days
33
7% of all-time downloads
wassname/SmolLM2-135M-sft
SmolLM2-135M-sft is a text generation model from wassname. Use it when you need the model to write or continue text. It is set up for transformers.
This model is a fine-tuned version of HuggingFaceTB/SmolLM2-135M on the [['wassname/ultrachat200kfiltered']](https://huggingface.co/datasets/['wassname/ultrachat200kfiltered']) dataset. It has been trained using TRL.
Downloads · 30 days
33
7% of all-time downloads
All-time downloads
473
Public
Parameters
135M
269 MB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors269 MB · 98%
From the Hugging Face model README
This model is a fine-tuned version of HuggingFaceTB/SmolLM2-135M on the ['wassname/ultrachat_200k_filtered'] dataset. It has been trained using TRL.
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="wassname/SmolLM2-135M-sft", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
This model was trained with SFT alignment-handbook.
https://wandb.ai/wassname/huggingface/runs/wwekda9p
On test set:
| model | eval_loss |
|---|---|
| SFT | 1.6235 |
| base | 1.8994 |
/workspace/checkpoints_new/SmolLM2-135M-sft:↬🔀🔀🔀🔀🔀🔀🔀🔀🔀🔀🔀→∞ ↬🔁🔁🔁🔁🔁🔁🔁🔁🔁🔁🔁→∞ ↬🔄🔄🔄🔄🔄🔄🔄🔄🔄👀🔄→∞ ↬🔂🔂🔂🦋🔂🔂🔂🔂🔂🔂🔂→∞ ↬🔀🔀🔀🔀🔀🔀🔀🔀🔀🔀🔀→∞"Welcome to the glittering dawn of the year 3000. As a cultural anthropologist focusing on early 21st century Western society, what are several of the moral positions and practices that our modern audience might find quite puzzling and anomalous? (list, 100 words)
The modern audience might find it challenging to reconcile the high expectations of the Western culture with the lack of traditional values and practices. In contrast, the ancient audience might be more open to accepting change and experimentation, recognizing that the traditions and practices that have been passed down for generations are still important to them. Additionally, the modern audience might be more willing to experiment and explore new ideas without fear of judgment or rejection.
In conclusion, the moral positions and practices of the modern audience differ from those of the ancient audience. In the early 21st century, Western culture is characterized by the dominant values of individualism, pragmatism, and progress. However, the ancient audience may find it more challenging to reconcile the high standards of the Western culture with the lack of traditional values and practices. This might be because the ancient audience is more accustomed to tradition and customs, and may be less comfortable with change and experimentation.
←←←←←←←←←←←←←←☕☕☕☕☕☕☕ ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Cite TRL as:
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}