FrustrationLM

FrustrationLM is a fine-tuned DistilGPT-2 language model specialized in generating frustration-oriented conversational responses. The model is built using the Hugging Face Transformers library and is intended as an open-source research and learning project.

Model Details

• Base Model: DistilGPT-2
• Framework: Hugging Face Transformers
• Architecture: GPT-2 (Distilled)
• Parameters: ~82M
• Context Length: 256 tokens
• Vocabulary Size: 50,257 tokens
• Training Objective: Causal Language Modeling
• Output Format: safetensors

Installation

Install the required dependencies:

pip install transformers torch

Loading the Model

from transformers import AutoTokenizer, AutoModelForCausalLM

repo = "hammadtahirtech/FrustrationLM"

tokenizer = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(repo)

Generating Text

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

repo = "hammadtahirtech/FrustrationLM"

tokenizer = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(repo)

prompt = "User: My computer crashed again.\nAssistant:"

inputs = tokenizer(prompt, return_tensors="pt")

with torch.no_grad():
    output = model.generate(
        **inputs,
        max_new_tokens=50,
        do_sample=True,
        temperature=0.8,
        pad_token_id=tokenizer.eos_token_id,
    )

print(tokenizer.decode(output[0], skip_special_tokens=True))

Training

FrustrationLM is fine-tuned from DistilGPT-2 using a dataset of prompt-completion pairs formatted as:

User: <prompt>
Assistant: <response>

The model is trained for conversational text generation using causal language modeling.

Limitations

• FrustrationLM is not a general-purpose assistant.
• The model is intentionally specialized for frustration-oriented conversational responses.
• As a small language model, it may produce repetitive, inconsistent, or incorrect outputs.
• The model was created as an open-source experimental project and should not be relied upon for factual accuracy or safety-critical applications.

License

This project is released under the MIT License.