DIIZZY/career-coach-v2

Career Coach - Fine-tuned Gemma 3 4B

This is a fine-tuned version of Gemma 3 4B specialized for career coaching and professional development advice.

⚠️ Note: This model was created using Unsloth with LoRA fine-tuning. It may require specific loading procedures.

Model Details

• Base Model: unsloth/gemma-3-4b-it-unsloth-bnb-4bit
• Fine-tuning Method: LoRA (Low-Rank Adaptation)
• Training Framework: Unsloth
• Model Size: ~4.3B parameters
• Specialization: Career coaching, professional development, job search advice

Usage

Option 1: Using Unsloth (Recommended)

from unsloth import FastLanguageModel

model, tokenizer = FastLanguageModel.from_pretrained(
    "DIIZZY/career-coach-v2",
    max_seq_length = 2048,
    dtype = None,
    load_in_4bit = True,
)

# Generate career advice
prompt = "What skills should I focus on to become a software engineer?"
inputs = tokenizer([prompt], return_tensors="pt")
outputs = model.generate(**inputs, max_length=200, temperature=0.7)
advice = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(advice)

Option 2: Standard Transformers (May require troubleshooting)

from transformers import AutoModelForCausalLM, AutoTokenizer

try:
    model = AutoModelForCausalLM.from_pretrained(
        "DIIZZY/career-coach-v2",
        trust_remote_code=True,
        torch_dtype="auto"
    )
    tokenizer = AutoTokenizer.from_pretrained("DIIZZY/career-coach-v2")
except Exception as e:
    print(f"Loading error: {e}")
    # Use Unsloth method instead

Training Data

The model was fine-tuned on career coaching conversations and professional development scenarios.

Intended Use

• Career transition guidance
• Skill development recommendations
• Interview preparation advice
• Professional growth strategies
• Job search optimization

Limitations

• Specialized for career-related topics
• May require Unsloth for proper loading
• Training data focused on tech/professional careers
• Model architecture may have compatibility issues with standard transformers

Troubleshooting

If you encounter loading issues:

1. Try using Unsloth's FastLanguageModel.from_pretrained()
2. Use trust_remote_code=True when loading
3. Check that you have the latest version of transformers and unsloth

License

Apache 2.0