README.md

---
base_model: unsloth/Llama-3.2-1B-bnb-4bit
language:
- en
license: apache-2.0
tags:
- text-generation-inference
- transformers
- unsloth
- llama
- trl
---

<div class="alert alert-block alert-success">  
<h2><center><strong>Mental Health Chatbot using 1B finetuned Llama 3.2 Model</strong></center></h2>
        
</div>

## Inference

```python
from unsloth import FastLanguageModel
model, tokenizer = FastLanguageModel.from_pretrained(
model_name = "ImranzamanML/1B_finetuned_llama3.2",
max_seq_length = 5020,
dtype = None,
load_in_4bit = True)
```


## Using this text to feed into model for getting the response 

text="I'm going through some things with my feelings and myself. I barely sleep and I do nothing but think about how I'm worthless and how I shouldn't be here. I've never tried or contemplated suicide. I've always wanted to fix my issues, but I never get around to it. How can I change my feeling of being worthless to everyone?"

<div style="background-color: #f2f2f2; border-left: 5px solid #4CAF50; padding: 15px; margin: 20px 0;">
    <strong>Note:</strong> Lets use the fine-tuned model for inference in order to generate responses based on mental health-related prompts !
</div>

<h3 style="color: #388e3c; font-family: Arial, sans-serif;">Here is some keys to note:</h3>

<ol style="margin-left: 20px;">
        <p>The <code>model = FastLanguageModel.for_inference(model)</code> configures the model specifically for inference, optimizing its performance for generating responses.</p>
    </li>
        <p>The input text is tokenized using the <code>tokenizer</code>, it convert the text into a format that model can process. We are using <code>data_prompt</code> to format the input text, while the response placeholder is left empty to get response from model. The <code>return_tensors = "pt"</code> parameter specifies that the output should be in PyTorch tensors, which are then moved to the GPU using <code>.to("cuda")</code> for faster processing.</p>
    </li>
        <p>The <code>model.generate</code> method generating response based on the tokenized inputs. The parameters <code>max_new_tokens = 5020</code> and <code>use_cache = True</code> ensure that the model can produce long and coherent responses efficiently by utilizing cached computation from previous layers.</p>
    </li>
</ol>

```python
model = FastLanguageModel.for_inference(model)
inputs = tokenizer(
[
    data_prompt.format(
        #instructions
        text,
        #answer
        "",
    )
], return_tensors = "pt").to("cuda")

outputs = model.generate(**inputs, max_new_tokens = 5020, use_cache = True)
answer=tokenizer.batch_decode(outputs)
answer = answer[0].split("### Response:")[-1]
print("Answer of the question is:", answer)
```