🇫🇷 Calme-3
Collection
Here you can find all the new Calme-3 models
•
27 items
•
Updated
•
9
This is avery small model, so it might not perform well for some prompts and may be sensitive to hyper parameters. I would appreciate any feedback to see if I can fix any issues in the next iteration. ❤️
This model is an advanced iteration of the powerful meta-llama/Llama-3.2-3B
, specifically fine-tuned to enhance its capabilities in French Legal domain.
All GGUF models are available here: MaziyarPanahi/calme-3.1-llamaloi-3b-GGUF
Detailed results can be found here
Metric | Value |
---|---|
Avg. | 24.01 |
IFEval (0-Shot) | 73.75 |
BBH (3-Shot) | 23.77 |
MATH Lvl 5 (4-Shot) | 16.77 |
GPQA (0-shot) | 4.14 |
MuSR (0-shot) | 1.11 |
MMLU-PRO (5-shot) | 24.50 |
This model uses ChatML
prompt template:
<|im_start|>system
{System}
<|im_end|>
<|im_start|>user
{User}
<|im_end|>
<|im_start|>assistant
{Assistant}
# Use a pipeline as a high-level helper
from transformers import pipeline
messages = [
{"role": "user", "content": "Who are you?"},
]
pipe = pipeline("text-generation", model="MaziyarPanahi/calme-3.1-llamaloi-3b")
pipe(messages)
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("MaziyarPanahi/calme-3.1-llamaloi-3b")
model = AutoModelForCausalLM.from_pretrained("MaziyarPanahi/calme-3.1-llamaloi-3b")
As with any large language model, users should be aware of potential biases and limitations. We recommend implementing appropriate safeguards and human oversight when deploying this model in production environments.