README.md

---
tags:
- seb
language:
- da
- no
- nb
- sv
license: mit
---

## Munin 7b e5
This model has 32 layers and the embedding size is 4096.

This model utilizes the lora adapter layer introduced in the paper [Improving Text Embeddings with Large Language Models](https://arxiv.org/pdf/2401.00368.pdf) along with the [merged model](https://huggingface.co/RJuro/munin-neuralbeagle-7b) by Roman Jurowetzki which merged the [Danish Munin model](https://huggingface.co/danish-foundation-models/munin-7b-alpha) with the [NeuralBeagle](https://huggingface.co/mlabonne/NeuralBeagle14-7B) model.


## Usage

### Loading the model

```python
from peft import PeftConfig, PeftModel
from transformers import AutoTokenizer, AutoModel

repo_id = "KennethEnevoldsen/munin-7b-e5"
config = PeftConfig.from_pretrained(repo_id)

base_model = AutoModel.from_pretrained(config.base_model_name_or_path)
model = PeftModel.from_pretrained(base_model, repo_id)
tokenizer = AutoTokenizer.from_pretrained(config.base_model_name_or_path)
```

Below is an example to encode queries and passages from the MS-MARCO passage ranking dataset.

```python
import torch
import torch.nn.functional as F

from torch import Tensor
from transformers import AutoTokenizer, AutoModel


def last_token_pool(last_hidden_states: Tensor,
                 attention_mask: Tensor) -> Tensor:
    left_padding = (attention_mask[:, -1].sum() == attention_mask.shape[0])
    if left_padding:
        return last_hidden_states[:, -1]
    else:
        sequence_lengths = attention_mask.sum(dim=1) - 1
        batch_size = last_hidden_states.shape[0]
        return last_hidden_states[torch.arange(batch_size, device=last_hidden_states.device), sequence_lengths]


def get_detailed_instruct(task_description: str, query: str) -> str:
    return f'Instruct: {task_description}\nQuery: {query}'


# Each query must come with a one-sentence instruction that describes the task
task = 'Given a web search query, retrieve relevant passages that answer the query'
queries = [
    get_detailed_instruct(task, 'how much protein should a female eat'),
    get_detailed_instruct(task, 'summit define')
]
# No need to add instruction for retrieval documents
documents = [
    "As a general guideline, the CDC's average requirement of protein for women ages 19 to 70 is 46 grams per day. But, as you can see from this chart, you'll need to increase that if you're expecting or training for a marathon. Check out the chart below to see how much protein you should be eating each day.",
    "Definition of summit for English Language Learners. : 1  the highest point of a mountain : the top of a mountain. : 2  the highest level. : 3  a meeting or series of meetings between the leaders of two or more governments."
]
input_texts = queries + documents

max_length = 4096
# Tokenize the input texts
batch_dict = tokenizer(input_texts, max_length=max_length - 1, return_attention_mask=False, padding=False, truncation=True)
# append eos_token_id to every input_ids
batch_dict['input_ids'] = [input_ids + [tokenizer.eos_token_id] for input_ids in batch_dict['input_ids']]
batch_dict = tokenizer.pad(batch_dict, padding=True, return_attention_mask=True, return_tensors='pt')

outputs = model(**batch_dict)
embeddings = last_token_pool(outputs.last_hidden_state, batch_dict['attention_mask'])

# normalize embeddings
embeddings = F.normalize(embeddings, p=2, dim=1)
scores = (embeddings[:2] @ embeddings[2:].T) * 100
print(scores.tolist())
```

## Supported Languages

This models is intended for use in Danish and Scandinavian languages.

## Evaluation
The model has not yet been evaluated. However we plan to evaluate it on [SEB](https://kennethenevoldsen.github.io/scandinavian-embedding-benchmark/)