Views
No views yet
gemma-4 designed to map candidate skills, job descriptions, and professional qualifications to official ESCO v1.2.1 (European Skills, Competences, Qualifications and Occupations) taxonomies and ISCO-08 unit codes.| Format | Repository | Recommended Use Case |
|---|---|---|
| 16-bit Safetensors | mazafard/esco-gemma4 | vLLM, transformers, cloud API servers, Python |
| GGUF Multi-Tier | mazafard/esco-gemma4-GGUF | Ollama, LM Studio, llama.cpp (q4_k_m, q8_0, f16) |
| Apple Silicon MLX | mazafard/esco-gemma4-MLX | Native macOS Apple Silicon Metal unified memory |
1Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.
2
3### Instruction:
4Map the following professional skills and experience to the correct ESCO occupation title and ISCO-08 code.
5
6### Input:
7<ENTER_CANDIDATE_SKILLS_HERE>
8
9### Response:1from unsloth import FastLanguageModel
2
3# 1. Load fine-tuned 16-bit or 4-bit model
4model, tokenizer = FastLanguageModel.from_pretrained(
5 model_name="mazafard/esco-gemma4",
6 max_seq_length=2048,
7 dtype=None,
8 load_in_4bit=True,
9)
10FastLanguageModel.for_inference(model)
11
12# 2. Define the Alpaca prompt template
13alpaca_prompt = """Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.
14
15### Instruction:
16Map the following professional skills and experience to the correct ESCO occupation title and ISCO-08 code.
17
18### Input:
19{}
20
21### Response:
22"""
23
24# 3. Format input skills
25skills = "Kubernetes, Docker container orchestration, Terraform IaC, CI/CD pipeline automation, Python, Prometheus monitoring, AWS cloud infrastructure"
26
27inputs = tokenizer(
28 [alpaca_prompt.format(skills)],
29 return_tensors="pt"
30).to("cuda")
31
32# 4. Generate deterministic output
33outputs = model.generate(
34 **inputs,
35 max_new_tokens=64,
36 use_cache=True,
37 temperature=0.1,
38 do_sample=False
39)
40
41decoded = tokenizer.decode(outputs[0], skip_special_tokens=True)
42response = decoded.split("### Response:")[-1].strip()
43print("🎯 ESCO Classification Result:
44", response)1ESCO Occupation Title: Cloud Engineer / DevOps Specialist
2ISCO-08 Code: 2512ollama run hf.co/mazafard/esco-gemma4-GGUF:q4_k_m1mlx_lm.generate \
2 --model mazafard/esco-gemma4-MLX \
3 --prompt "### Instruction:\nMap the following professional skills and experience to the correct ESCO occupation title and ISCO-08 code.\n\n### Input:\nKubernetes, Docker, CI/CD\n\n### Response:\n" \
4 --max-tokens 641@misc{esco_taxonomy_2024,
2 author = {{European Commission}},
3 title = {European Skills, Competences, Qualifications and Occupations (ESCO) Dataset v1.2.1},
4 year = {2024},
5 publisher = {European Union Portal},
6 howpublished = {\url{https://esco.ec.europa.eu/}},
7 note = {Accessed: 2026-05-26}
8}
9
10@software{gemma4_esco_finetune_2026,
11 author = {Fard, Mohammadreza A.},
12 title = {Parameter-Efficient Fine-Tuning (PEFT) and Telemetry Pipeline for Gemma-4 on ESCO Skill Inventories},
13 month = may,
14 year = {2026},
15 publisher = {GitHub Repository},
16 version = {1.0.0},
17 url = {https://github.com/mazafard/esco-gemma4-pipeline}
18}