40 lines
1.6 KiB
Markdown
40 lines
1.6 KiB
Markdown
---
|
|
datasets:
|
|
- k10shetty/resume-skill-extractor-dataset
|
|
base_model: meta-llama/Meta-Llama-3-8B-Instruct
|
|
library_name: transformers
|
|
tags:
|
|
- text-generation
|
|
- fine-tuning
|
|
- llama-3
|
|
- nlp
|
|
- resume-parsing
|
|
- job-description
|
|
license: llama3
|
|
language:
|
|
- en
|
|
---
|
|
|
|
# 🚀 Resume Skill Extractor (Llama 3 8B)
|
|
|
|
## Model Overview
|
|
This is a fully merged, standalone 8-Billion parameter language model specifically fine-tuned to act as an expert technical recruiter. It takes raw, unstructured job descriptions and instantly extracts a concise summary along with a cleanly formatted, bulleted list of mandatory and bonus technical skills.
|
|
|
|
This model was created by fine-tuning `Meta-Llama-3-8B-Instruct` using QLoRA, and the resulting adapters have been permanently merged into the base weights for simplified deployment and serverless inference.
|
|
|
|
* **Base Model:** Meta-Llama-3-8B-Instruct
|
|
* **Original LoRA Adapters:** [k10shetty/resume-skill-extractor-lora](https://huggingface.co/k10shetty/resume-skill-extractor-lora)
|
|
* **Task:** Information Extraction / Text Generation
|
|
* **Architecture:** LlamaForCausalLM (16-bit bfloat16)
|
|
|
|
## Prompt Format
|
|
This model retains the standard Llama 3 Instruct prompt format. To get the best results, you **must** use this exact structure:
|
|
|
|
```text
|
|
<|begin_of_text|><|start_header_id|>system<|end_header_id|>
|
|
You are an expert technical recruiter. Analyze job descriptions and extract key information.<|eot_id|>
|
|
<|start_header_id|>user<|end_header_id|>
|
|
Analyze this job description and provide a summary, the required skills, and the bonus/nice-to-have skills:
|
|
|
|
[INSERT JOB DESCRIPTION HERE]<|eot_id|>
|
|
<|start_header_id|>assistant<|end_header_id|> |