---
library_name: peft
---

## Inference

```python
from transformers import AutoTokenizer
import transformers
import torch

model = "qanastek/LLaMa-2-FrenchMedMCQA"

tokenizer = AutoTokenizer.from_pretrained(model)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

prompt = "Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request. ### Instruction: We are giving you a scientific question (easy level) and five answers options (associated to « A », « B », « C », « D », « E »). Your task is to find the correct(s) answer(s) based on scientific facts, knowledge and reasoning. Don't generate anything other than one of the following characters : 'A B C D E'. ### Input: Parmi les propositions suivantes, quelle est celle qui est exacte? Lorsqu'on ajoute un acide fort à une solution tampon: (A) Le pH reste constant (B) Le pH diminue légèrement (C) Le constituant basique du tampon reste constant (D) Le constituant acide du tampon réagit (E) Le rapport acide/base reste inchangé ### Response: "

seq = pipeline(
    prompt,
    do_sample=True,
    top_k=10,
    num_return_sequences=1,
    eos_token_id=tokenizer.eos_token_id,
    max_length=200,
)[0]

print(seq['generated_text'][len(prompt):])
```

## Training procedure


The following `bitsandbytes` quantization config was used during training:
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: False
- bnb_4bit_compute_dtype: float16

The following `bitsandbytes` quantization config was used during training:
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: False
- bnb_4bit_compute_dtype: float16
### Framework versions

- PEFT 0.4.0

- PEFT 0.4.0