license: apache-2.0
tags:
- mixtral
- finetune
model-index:
- name: Smaug-Mixtral-v0.1
results:
- task:
type: text-generation
name: Text Generation
dataset:
name: AI2 Reasoning Challenge (25-Shot)
type: ai2_arc
config: ARC-Challenge
split: test
args:
num_few_shot: 25
metrics:
- type: acc_norm
value: 74.91
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=abacusai/Smaug-Mixtral-v0.1
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: HellaSwag (10-Shot)
type: hellaswag
split: validation
args:
num_few_shot: 10
metrics:
- type: acc_norm
value: 87.79
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=abacusai/Smaug-Mixtral-v0.1
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MMLU (5-Shot)
type: cais/mmlu
config: all
split: test
args:
num_few_shot: 5
metrics:
- type: acc
value: 70.08
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=abacusai/Smaug-Mixtral-v0.1
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: TruthfulQA (0-shot)
type: truthful_qa
config: multiple_choice
split: validation
args:
num_few_shot: 0
metrics:
- type: mc2
value: 66.88
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=abacusai/Smaug-Mixtral-v0.1
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: Winogrande (5-shot)
type: winogrande
config: winogrande_xl
split: validation
args:
num_few_shot: 5
metrics:
- type: acc
value: 81.69
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=abacusai/Smaug-Mixtral-v0.1
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: GSM8k (5-shot)
type: gsm8k
config: main
split: test
args:
num_few_shot: 5
metrics:
- type: acc
value: 72.48
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=abacusai/Smaug-Mixtral-v0.1
name: Open LLM Leaderboard
Overview
This model is part of the Smaug series of finetuned models. This one based on https://huggingface.co/mistralai/Mixtral-8x7B-v0.1
We use a new fine-tuning technique, DPO-Positive (DPOP), and new pairwise preference versions of ARC, HellaSwag, and MetaMath (as well as other existing datasets). We introduce the technique and the full training details in our new paper: https://arxiv.org/abs/2402.13228.
We show that on datasets in which the edit distance between pairs of completions is low (such as in math-based datasets), standard DPO loss can lead to a reduction of the model's likelihood of the preferred examples, as long as the relative probability between the preferred and dispreferred classes increases. Using these insights, we design DPOP, a new loss function and training procedure which avoids this failure mode. Surprisingly, we also find that DPOP outperforms DPO across a wide variety of datasets and downstream tasks, including datasets with high edit distances between completions.
We believe this new approach is generally useful in training across a wide range of model types and downstream use cases, and it powers all of our Smaug models. With the release of our paper and datasets, we are excited for the open source community to continue to build on and improve Smaug and spawn more dragons to dominate the LLM space!
Keep watching this space for our announcements!
Evaluation Results
| Average | ARC | HellaSwag | MMLU | TruthfulQA | Winogrande | GSM8K |
|---|---|---|---|---|---|---|
| 75.12 | 74.91 | 87.70 | 70.16 | 65.96 | 81.61 | 70.36 |
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 75.64 |
| AI2 Reasoning Challenge (25-Shot) | 74.91 |
| HellaSwag (10-Shot) | 87.79 |
| MMLU (5-Shot) | 70.08 |
| TruthfulQA (0-shot) | 66.88 |
| Winogrande (5-shot) | 81.69 |
| GSM8k (5-shot) | 72.48 |