Mistral-Instruct-7B-v0.2-ChatAlpaca

Model description

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.2 on the English robinsmits/ChatAlpaca-20K dataset.

It achieves the following results on the evaluation set:

• Loss: 0.8584

Model usage

A basic example of how to use the finetuned model. Note this example is a modified version from the base model.

import torch
from peft import AutoPeftModelForCausalLM
from transformers import AutoTokenizer

device = "cuda"

model = AutoPeftModelForCausalLM.from_pretrained("robinsmits/Mistral-Instruct-7B-v0.2-ChatAlpaca", 
                                                 device_map = "auto", 
                                                 load_in_4bit = True, 
                                                 torch_dtype = torch.bfloat16)

tokenizer = AutoTokenizer.from_pretrained("robinsmits/Mistral-Instruct-7B-v0.2-ChatAlpaca")

messages = [
    {"role": "user", "content": "What is your favourite condiment?"},
    {"role": "assistant", "content": "Well, I'm quite partial to a good squeeze of fresh lemon juice. It adds just the right amount of zesty flavour to whatever I'm cooking up in the kitchen!"},
    {"role": "user", "content": "Do you have mayonnaise recipes?"}
]

encodeds = tokenizer.apply_chat_template(messages, return_tensors = "pt")

generated_ids = model.generate(input_ids = encodeds.to(device), max_new_tokens = 512, do_sample = True)
decoded = tokenizer.batch_decode(generated_ids)
print(decoded[0])

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 4e-05
• train_batch_size: 1
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 32
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss
0.99	0.2	120	0.9355
0.8793	0.39	240	0.8848
0.8671	0.59	360	0.8737
0.8662	0.78	480	0.8679
0.8627	0.98	600	0.8639
0.8426	1.18	720	0.8615
0.8574	1.37	840	0.8598
0.8473	1.57	960	0.8589
0.8528	1.76	1080	0.8585
0.852	1.96	1200	0.8584

Framework versions

• PEFT 0.7.1
• Transformers 4.36.2
• Pytorch 2.1.2+cu121
• Datasets 2.16.0
• Tokenizers 0.15.0

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	61.21
AI2 Reasoning Challenge (25-Shot)	56.74
HellaSwag (10-Shot)	80.82
MMLU (5-Shot)	59.10
TruthfulQA (0-shot)	55.86
Winogrande (5-shot)	77.11
GSM8k (5-shot)	37.60