ShieldX_-_manovyadh-1.1B-v1-chat-awq

Quantization made by Richard Erkhov.

Github

Discord

Request more models

manovyadh-1.1B-v1-chat - AWQ

• Model creator: https://huggingface.co/ShieldX/
• Original model: https://huggingface.co/ShieldX/manovyadh-1.1B-v1-chat/

Original model description:

library_name: transformers tags:

• medical
• trl
• trainer license: apache-2.0 thumbnail: https://huggingface.co/ShieldX/manovyadh-1.1B-v1-chat/blob/main/manovyadh.png datasets:
• ShieldX/manovyadh-3.5k language:
• en metrics:
• accuracy pipeline_tag: text-generation base_model: TinyLlama/TinyLlama-1.1B-Chat-v1.0 widget:

  • text: > ###SYSTEM: You are an AI assistant that helps people cope with stress and improve their mental health. User will tell you about their feelings and challenges. Your task is to listen empathetically and offer helpful suggestions. While responding, think about the user’s needs and goals and show compassion and support

    ###USER: I don't know how to tell someone how I feel about them. How can I get better at expressing how I feel??

    ###ASSISTANT: model-index:

  • name: manovyadh-1.1B-v1-chat results:

    • task: type: text-generation dataset: name: ai2_arc type: arc metrics:
      • name: pass@1 type: pass@1 value: 35.92 source: name: Open LLM Leaderboard url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard
    • task: type: text-generation dataset: name: hellaswag type: hellaswag metrics:
      • name: pass@1 type: pass@1 value: 60.03 source: name: Open LLM Leaderboard url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard
    • task: type: text-generation dataset: name: truthful_qa type: truthful_qa metrics:
      • name: pass@1 type: pass@1 value: 39.17 source: name: Open LLM Leaderboard url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard
    • task: type: text-generation dataset: name: winogrande type: winogrande metrics:
      • name: pass@1 type: pass@1 value: 61.09 source: name: Open LLM Leaderboard url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard

---

Uploaded model

• Developed by: ShieldX
• License: apache-2.0
• Finetuned from model : TinyLlama/TinyLlama-1.1B-Chat-v1.0

<style> img{ width: 40vw; height: auto; margin: 0 auto; display: flex; align-items: center; justify-content: center; } </style>

ShieldX/manovyadh-1.1B-v1

Introducing ManoVyadh, A finetuned version of TinyLlama 1.1B Chat on Mental Health Counselling Dataset.

<img class="custom-image" src="manovyadh.png" alt="BongLlama">

Model Details

Model Description

ManoVyadh is a LLM for mental health counselling.

Uses

Direct Use

• base model for further finetuning
• for fun

Downstream Use

• can be deployed with api
• used to create webapp or app to show demo

Out-of-Scope Use

• cannot be used for production purpose
• not to be applied in real life health purpose
• cannot be used to generate text for research or academic purposes

Usage

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM, AutoConfig

tokenizer = AutoTokenizer.from_pretrained("ShieldX/manovyadh-1.1B-v1-chat")
model = AutoModelForCausalLM.from_pretrained("ShieldX/manovyadh-1.1B-v1-chat").to("cuda")
config = AutoConfig.from_pretrained("ShieldX/manovyadh-1.1B-v1-chat")

def format_prompt(q):
  return f"""###SYSTEM: You are an AI assistant that helps people cope with stress and improve their mental health. User will tell you about their feelings and challenges. Your task is to listen empathetically and offer helpful suggestions. While responding, think about the user’s needs and goals and show compassion and support
      ###USER: {q}
      ###ASSISTANT:"""

prompt = format_prompt("I've never been able to talk with my parents. My parents are in their sixties while I am a teenager. I love both of them but not their personalities. I feel that they do not take me seriously whenever I talk about a serious event in my life. If my dad doesn’t believe me, then my mom goes along with my dad and acts like she doesn’t believe me either. I’m a pansexual, but I can’t trust my own parents. I've fought depression and won; however, stress and anxiety are killing me. I feel that my friends don't listen to me. I know they have their own problems, which I do my best to help with. But they don't always try to help me with mine, when I really need them. I feel as if my childhood has been taken from me. I feel as if I have no one whom I can trust.")

import torch
from transformers import GenerationConfig, TextStreamer
from time import perf_counter

# Check for GPU availability
if torch.cuda.is_available():
    device = "cuda"
else:
    device = "cpu"

# Move model and inputs to the GPU (if available)
model.to(device)

inputs = tokenizer(prompt, return_tensors="pt").to(device)

streamer = TextStreamer(tokenizer)

generation_config = GenerationConfig(
    penalty_alpha=0.6,
    do_sample=True,
    top_k=5,
    temperature=0.5,
    repetition_penalty=1.2,
    max_new_tokens=256,
    streamer=streamer,
    pad_token_id=tokenizer.eos_token_id
)

start_time = perf_counter()
outputs = model.generate(**inputs, generation_config=generation_config)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
output_time = perf_counter() - start_time
print(f"Time taken for inference: {round(output_time, 2)} seconds")

Bias, Risks, and Limitations

Significant research has explored bias and fairness issues with language models (see, e.g., Sheng et al. (2021) and Bender et al. (2021)). Predictions generated by the model may include disturbing and harmful stereotypes across protected classes; identity characteristics; and sensitive, social, and occupational groups.

Training Details

Model Examination

We will be further finetuning this model on large dataset to see how it performs

Environmental Impact

Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

• Hardware Type: 1 X Tesla T4
• Hours used: 0.48
• Cloud Provider: Google Colab
• Compute Region: India

Technical Specifications

Model Architecture and Objective

Finetuned on Tiny-Llama 1.1B Chat model

Hardware

1 X Tesla T4

training

This model is a fine-tuned version of TinyLlama/TinyLlama-1.1B-Chat-v1.0 on ShieldX/manovyadh-3.5k dataset. It achieves the following results on the evaluation set:

• Loss: 1.8587

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2.5e-05
• train_batch_size: 2
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 400
• mixed_precision_training: Native AMP

Training results

Framework versions

• PEFT 0.7.1
• Transformers 4.37.1
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.1

Citation

<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->

BibTeX:

@misc{ShieldX/manovyadh-1.1B-v1-chat,
      url={[https://huggingface.co/ShieldX/manovyadh-1.1B-v1-chat](https://huggingface.co/ShieldX/manovyadh-1.1B-v1-chat)},
      title={ManoVyadh},
      author={Rohan Shaw},
      year={2024}, month={Jan}
}

Model Card Authors

ShieldX a.k.a Rohan Shaw

Model Card Contact

email : [email protected]