File size: 5,625 Bytes
728605f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 |
---
license: apache-2.0
base_model: HuggingFaceTB/SmolLM-360M
tags:
- alignment-handbook
- trl
- sft
datasets:
- Magpie-Align/Magpie-Pro-300K-Filtered
- bigcode/self-oss-instruct-sc2-exec-filter-50k
- teknium/OpenHermes-2.5
- HuggingFaceTB/everyday-conversations-llama3.1-2k
library_name: transformers
language:
- en
---
# QuantFactory/smollm-360M-instruct-add-basics-GGUF
This is quantized version of [HuggingFaceTB/smollm-360M-instruct-add-basics](https://huggingface.co/HuggingFaceTB/smollm-360M-instruct-add-basics) created using llama.cpp
# Original Model Card
# SmolLM-360M-Instruct
<center>
<img src="https://huggingface.co/datasets/HuggingFaceTB/images/resolve/main/banner_smol.png" alt="SmolLM" width="1100" height="600">
</center>
## Model Summary
Chat with the model at: https://huggingface.co/spaces/HuggingFaceTB/instant-smol
SmolLM is a series of language models available in three sizes: 135M, 360M, and 1.7B parameters.
These models are trained on [SmolLM-Corpus](https://huggingface.co/datasets/HuggingFaceTB/smollm-corpus), a curated collection of high-quality educational and synthetic data designed for training LLMs. For further details, we refer to our [blogpost](https://huggingface.co/blog/smollm).
To build SmolLM-Instruct, we finetune the base models on publicly available datasets.
## Changelog
|Release|Description|
|-|-|
|v0.1| Initial release of SmolLM-Instruct. We finetune on the permissive subset of the WebInstructSub dataset, combined with StarCoder2-Self-OSS-Instruct. Then, we perform DPO (Direct Preference Optimization) for one epoch on HelpSteer for the 135M and 1.7B models, and argilla/dpo-mix-7k for the 360M model.|
|v0.2| We changed the finetuning mix to datasets more suitable for smol models. We train on a new dataset of 2k simple everyday conversations we generated by llama3.1-70B [everyday-conversations-llama3.1-2k](https://huggingface.co/datasets/HuggingFaceTB/everyday-conversations-llama3.1-2k/), [Magpie-Pro-300K-Filtere](https://huggingface.co/datasets/Magpie-Align/Magpie-Pro-300K-Filtered), [self-oss-instruct-sc2-exec-filter-50k](https://huggingface.co/datasets/bigcode/self-oss-instruct-sc2-exec-filter-50k), and a small subset of [OpenHermes-2.5](https://huggingface.co/datasets/teknium/OpenHermes-2.5)|
We've noticed that the v0.2 models are better at staying on topic and responding appropriately to standard prompts, such as greetings and questions about their role as AI assistants. Additionally, SmolLM-360M-Instruct (v0.2) has a 63.3% win rate over SmolLM-360M-Instruct (v0.1) on AlpacaEval. You can find the details [here](https://huggingface.co/datasets/HuggingFaceTB/alpaca_eval_details/).
## Usage
### Local Applications
⚡ For local applications, you can find optimized implementations of the model in MLC, GGUF and Transformers.js formats, in addition to fast in-browser demos in this collection: https://huggingface.co/collections/HuggingFaceTB/local-smollms-66c0f3b2a15b4eed7fb198d0
We noticed that 4bit quantization degrades the quality of the 135M and 360M, so we use `q016` for MLC and ONNX/Transformers.js checkpoints for the WebGPU demos. We also suggest using temperature 0.2 and top-p 0.9.
### Transformers
```bash
pip install transformers
```
```python
# pip install transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
checkpoint = "HuggingFaceTB/SmolLM-360M-Instruct"
device = "cuda" # for GPU usage or "cpu" for CPU usage
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
# for multiple GPUs install accelerate and do `model = AutoModelForCausalLM.from_pretrained(checkpoint, device_map="auto")`
model = AutoModelForCausalLM.from_pretrained(checkpoint).to(device)
messages = [{"role": "user", "content": "What is the capital of France."}]
input_text=tokenizer.apply_chat_template(messages, tokenize=False)
print(input_text)
inputs = tokenizer.encode(input_text, return_tensors="pt").to(device)
outputs = model.generate(inputs, max_new_tokens=50, temperature=0.2, top_p=0.9, do_sample=True)
print(tokenizer.decode(outputs[0]))
```
### Chat in TRL
You can also use the TRL CLI to chat with the model from the terminal:
```bash
pip install trl
trl chat --model_name_or_path HuggingFaceTB/SmolLM-360M-Instruct --device cpu
```
## Limitations
Additionally, the generated content may not always be factually accurate, logically consistent, or free from biases present in the training data, we invite users to leverage them as assistive tools rather than definitive sources of information. We find that they can handle general knowledge questions, creative writing and basic Python programming. But they are English only and may have difficulty with arithmetics, editing tasks and complex reasoning. For more details about the models' capabilities, please refer to our [blog post](https://huggingface.co/blog/smollm).
## Training parameters
We train the models using the [alignement-handbook](https://github.com/huggingface/alignment-handbook) with the datasets mentioned in the changelog, using these parameters for v0.2:
- 1 epoch
- lr 1e-3
- cosine schedule
- warmup ratio 0.1
- global batch size 262k tokens
# Citation
```bash
@misc{allal2024SmolLM,
title={SmolLM - blazingly fast and remarkably powerful},
author={Loubna Ben Allal and Anton Lozhkov and Elie Bakouch and Leandro von Werra and Thomas Wolf},
year={2024},
}
```
|