README.md

---
license: apache-2.0
model-index:
- name: Graphcore/gptj-mnli
  results: []
datasets:
- multi_nli
tags:
- pytorch
- causal-lm
- text-generation
widget:
- text: "mnli hypothesis: Your contributions were of no help with our students' education. premise: Your contribution helped make it possible for us to provide our students with a quality education. target:"
---

# Graphcore/gptj-mnli

This model is the fine-tuned version of [EleutherAI/gpt-j-6B](https://huggingface.co/EleutherAI/gpt-j-6B) on the [MNLI dataset]()

MNLI dataset consists of pairs of sentences, a *premise* and a *hypothesis*.
The task is to predict the relation between the premise and the hypothesis, which can be:
- `entailment`: hypothesis follows from the premise,
- `contradiction`: hypothesis contradicts the premise,
- `neutral`: hypothesis and premise are unrelated.

The MNLI task is to take two sentences referred to as the hypothesis and the premise as input and decide if the sentences entail (support), are neutral (cover different subjects) or contradict each other.

We finetune the model as a Causal Language Model (CLM): given a sequence of tokens, the task is to predict the next token.
To achieve this, we create a stylised prompt string, following the approach of [T5 paper](https://arxiv.org/pdf/1910.10683.pdf).
```shell
mnli hypothesis: {hypothesis} premise: {premise} target: {class_label} <|endoftext|>
```
For example:
```
mnli hypothesis: Your contributions were of no help with our students' education. premise: Your contribution helped make it possible for us to provide our students with a quality education. target: contradiction <|endoftext|>
```

## Fine-tuning and validation data
Fine tuning is done using the `train` split of the GLUE MNLI dataset and the performance is measured using the `validation_mismatched` split.

`validation_mismatched` means validation examples are not derived from the same sources as those in the training set and therefore not closely resembling any of the examples seen at training time.

## Fine-tuning procedure
Fine tuned on a Graphcore IPU-POD64 using `popxl`.     

Prompt sentences are tokenized and packed together to form 1024 token sequences, following [HF packing algorithm](https://github.com/huggingface/transformers/blob/v4.20.1/examples/pytorch/language-modeling/run_clm.py). No padding is used.
Since the model is trained to predict the next token, labels are simply the input sequence shifted by one token.
Given the training format, no extra care is needed to account for different sequences: the model does not need to know which sentence a token belongs to.

## How to use
The model can be easily loaded using AutoModelForCausalLM.
Text generation can be implemented or via the pipeline API.

```python
from transformers import pipeline, AutoModelForCausalLM, AutoTokenizer

hf_model = AutoModelForCausalLM.from_pretrained("Graphcore/gptj-mnli")
tokenizer = AutoTokenizer.from_pretrained('EleutherAI/gpt-j-6B')
generator =  pipeline('text-generation', model=hf_model, tokenizer=tokenizer)
prompt = "mnli hypothesis: Your contributions were of no help with our students' education." \
         "premise: Your contribution helped make it possible for us to provide our students with a quality education. target:"
out = generator(prompt, return_full_text=False, max_new_tokens=5, top_k=1)
# [{'generated_text': ' contradiction'}]
```