metadata
language:
- en
license: llama3.2
tags:
- text-generation-inference
- transformers
- unsloth
- llama
- llama-3
- trl
- sft
base_model: unsloth/Llama-3.2-1B-Instruct-bnb-4bit
datasets:
- mlabonne/FineTome-100k
model-index:
- name: FineTome-Llama3.2-1B-0929
results:
- task:
type: text-generation
name: Text Generation
dataset:
name: IFEval (0-Shot)
type: HuggingFaceH4/ifeval
args:
num_few_shot: 0
metrics:
- type: inst_level_strict_acc and prompt_level_strict_acc
value: 39.91
name: strict accuracy
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=NotASI/FineTome-Llama3.2-1B-0929
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: BBH (3-Shot)
type: BBH
args:
num_few_shot: 3
metrics:
- type: acc_norm
value: 5.74
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=NotASI/FineTome-Llama3.2-1B-0929
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MATH Lvl 5 (4-Shot)
type: hendrycks/competition_math
args:
num_few_shot: 4
metrics:
- type: exact_match
value: 1.28
name: exact match
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=NotASI/FineTome-Llama3.2-1B-0929
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: GPQA (0-shot)
type: Idavidrein/gpqa
args:
num_few_shot: 0
metrics:
- type: acc_norm
value: 3.02
name: acc_norm
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=NotASI/FineTome-Llama3.2-1B-0929
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MuSR (0-shot)
type: TAUR-Lab/MuSR
args:
num_few_shot: 0
metrics:
- type: acc_norm
value: 2.66
name: acc_norm
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=NotASI/FineTome-Llama3.2-1B-0929
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MMLU-PRO (5-shot)
type: TIGER-Lab/MMLU-Pro
config: main
split: test
args:
num_few_shot: 5
metrics:
- type: acc
value: 4.76
name: accuracy
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=NotASI/FineTome-Llama3.2-1B-0929
name: Open LLM Leaderboard
Notice
Model was submitted to OpenLLM Leaderboard for full evaluation.
- MMLU-PRO (5-shot) (self-reported): 0.1553 ± 0.0033
- MMLU (0-shot) (self-reported): 0.3416 ± 0.0040
- Hellaswag (0-shot) (self-reported):
- acc: 0.4284 ± 0.0049
- acc_norm: 0.5681 ± 0.0049
Code + Math optimized version coming soon!
IMPORTANT
In case you got the following error:
exception: data did not match any variant of untagged enum modelwrapper at line 1251003 column 3
Please upgrade your transformer package, that is, use the following code:
pip install --upgrade "transformers>=4.45"
Uploaded model
- Developed by: NotASI
- License: apache-2.0
- Finetuned from model : unsloth/Llama-3.2-1B-Instruct-bnb-4bit
Details
This model was trained on mlabonne/FineTome-100k for 2 epochs with rslora + qlora, and achieve the final training loss: 0.796700.
This model follows the same chat template as the base model one.
This llama model was trained 2x faster with Unsloth and Huggingface's TRL library.
Usage
Sample notebook (Unsloth): https://colab.research.google.com/drive/1cSCvemKsC0JVmMs4jdfoTqM8SQQ_DSBt?usp=sharing
Dependencies
!pip install unsloth
!pip install --upgrade "transformers>=4.45"
Code
from unsloth.chat_templates import get_chat_template
from unsloth import FastLanguageModel
import torch
model, tokenizer = FastLanguageModel.from_pretrained(
model_name = "NotASI/FineTome-Llama3.2-1B-0929",
)
tokenizer = get_chat_template(
tokenizer,
chat_template = "llama-3.1",
)
FastLanguageModel.for_inference(model) # Enable native 2x faster inference
messages = [
{"role": "user", "content": "Hi!"},
]
inputs = tokenizer.apply_chat_template(
messages,
tokenize = True,
add_generation_prompt = True, # Must add for generation
return_tensors = "pt",
).to("cuda")
from transformers import TextStreamer
text_streamer = TextStreamer(tokenizer, skip_prompt = True)
_ = model.generate(input_ids = inputs, streamer = text_streamer, max_new_tokens = 128,
use_cache = True, temperature = 1.5, min_p = 0.1)
Example 1
Input:
Continue the fibonnaci sequence: 1, 1, 2, 3, 5, 8. Give me the next 10 numbers.
Output:
Here are the next 10 numbers in the Fibonacci sequence:
1. 13
2. 21
3. 34
4. 55
5. 89
6. 144
7. 233
8. 377
9. 610
10. 987
11. 1597
12. 2584
13. 4190
14. 6765
15. 10946
16. 17715
Explanation: The Fibonacci sequence is a series of numbers where each number is the sum of the two preceding ones, starting from 0 and 1. To calculate the nth Fibonacci number, you can use the formula F(n) = F(n-1) + F(n-2).<|eot_id|>
Example 2
Input:
Who are you? Briefy introduce yourself.
Output:
Hi there! I'm an AI assistant designed to help with a range of inquiries. As for your request to introduce yourself briefly, I can certainly do that. I am a highly advanced AI, capable of understanding and processing vast amounts of information. My purpose is to provide useful and informative responses to the users' questions.<|eot_id|>
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 9.56 |
| IFEval (0-Shot) | 39.91 |
| BBH (3-Shot) | 5.74 |
| MATH Lvl 5 (4-Shot) | 1.28 |
| GPQA (0-shot) | 3.02 |
| MuSR (0-shot) | 2.66 |
| MMLU-PRO (5-shot) | 4.76 |