Update README: remove Unsloth branding, add detailed capabilities and usage

README.md

@@ -1,22 +1,158 @@
 ---
-base_model: unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+base_model: google/gemma-4-e2b-it
 tags:
 - text-generation-inference
 - transformers
-- unsloth
 - gemma4
 - trl
+- peft
+- cybersecurity
+- devsecops
+- security
+- lora
 license: apache-2.0
 language:
 - en
 ---
 
-# Uploaded  model
+# Gemma 4 E2B — Cybersecurity Interview Expert
 
-- **Developed by:** rezaduty
-- **License:** apache-2.0
-- **Finetuned from model :** unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+A QLoRA fine-tuned version of [Gemma 4 E2B Instruct](https://huggingface.co/google/gemma-4-e2b-it) specialized in **deep, production-level cybersecurity knowledge**. This model answers technical security interview questions with precision, concrete examples, and actionable recommendations.
 
-This gemma4 model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth)
+---
+
+## Model Details
+
+| Property | Value |
+|---|---|
+| **Base model** | google/gemma-4-e2b-it (2B parameters) |
+| **Fine-tuning method** | QLoRA (rank 16, α 16) |
+| **Trainable parameters** | 31M / 5.15B (0.60%) |
+| **Training data** | 646 curated cybersecurity interview Q&A pairs |
+| **Epochs** | 3 |
+| **Final training loss** | 0.574 |
+| **License** | Apache 2.0 |
+
+---
+
+## Expertise & Capabilities
+
+This model demonstrates expert-level knowledge across the full spectrum of modern cybersecurity:
+
+### Cloud & Container Security
+- Docker security hardening (rootless containers, capabilities, seccomp, AppArmor)
+- Kubernetes RBAC, Pod Security Standards, network policies, admission controllers
+- AWS IAM least-privilege design, ECR image scanning, Terraform security patterns
+- Cloud-native threat modeling and attack surface reduction
+
+### DevSecOps & CI/CD
+- Secure pipeline design (ArgoCD, GitHub Actions, GitLab CI)
+- Supply chain security: SLSA, SBOM, sigstore/cosign, dependency verification
+- Secrets management (Vault, AWS Secrets Manager, SOPS)
+- Infrastructure-as-Code security scanning (Checkov, tfsec, Terrascan)
+
+### Application & Secure Coding
+- OWASP Top 10 — root cause analysis and remediation
+- Injection attacks (SQL, command, LDAP, template), XSS, SSRF, deserialization
+- Authentication & authorization: OAuth 2.0, OIDC, JWT, PKCE, session security
+- Cryptography: TLS configuration, key management, algorithm selection
+
+### Threat Intelligence & Offensive Security
+- SOC operations, SIEM correlation rules, threat hunting
+- MITRE ATT&CK mapping and adversary emulation
+- Active Directory attack paths (Kerberoasting, Pass-the-Hash, DCSync)
+- Red team tactics and purple team collaboration
+
+### Emerging & Specialized Domains
+- AI/LLM security: prompt injection, model poisoning, guardrail bypasses
+- OT/ICS/SCADA security: Purdue model, IEC 62443, air-gap strategies
+- Blockchain & smart contract auditing (reentrancy, overflow, access control)
+- Digital forensics, incident response, and malware analysis
+
+---
+
+## Usage
+
+```python
+from transformers import AutoTokenizer, AutoModelForCausalLM
+from peft import PeftModel
+import torch
+
+base_model = "google/gemma-4-e2b-it"
+adapter    = "rezaduty/gemma4-e2b-cybersecurity-interview"
+
+tokenizer = AutoTokenizer.from_pretrained(adapter)
+model = AutoModelForCausalLM.from_pretrained(
+    base_model,
+    torch_dtype=torch.bfloat16,
+    device_map="auto",
+)
+model = PeftModel.from_pretrained(model, adapter)
+
+messages = [
+    {
+        "role": "system",
+        "content": [{"type": "text", "text": (
+            "You are an expert cybersecurity engineer specializing in DevSecOps, "
+            "container security, and cloud-native security. Answer technical interview "
+            "questions with depth, precision, and concrete examples."
+        )}]
+    },
+    {
+        "role": "user",
+        "content": [{"type": "text", "text": "Explain why running Docker containers as root is a security risk and how to fix it."}]
+    },
+]
+
+inputs = tokenizer.apply_chat_template(
+    messages,
+    tokenize=True,
+    add_generation_prompt=True,
+    return_tensors="pt",
+).to(model.device)
+
+output = model.generate(
+    input_ids=inputs,
+    max_new_tokens=512,
+    temperature=0.7,
+    top_p=0.9,
+    use_cache=True,
+)
+print(tokenizer.decode(output[0][inputs.shape[-1]:], skip_special_tokens=True))
+```
+
+---
+
+## Training Dataset
+
+Covers 15 curated topic domains across 646 high-quality question/answer pairs:
+
+- Container & Kubernetes security
+- Cloud IAM, ECR, Terraform security
+- CI/CD and ArgoCD pipeline security
+- AI/LLM security
+- DevOps patterns and security tooling
+- Secure coding (OWASP, injection, crypto)
+- SOC operations and threat intelligence
+- Active Directory and red team techniques
+- Software architecture and design security
+- Authentication, identity, and supply chain
+- OT/ICS/SCADA security
+- Blockchain and smart contract security
+- OS hardening, cloud SaaS, and forensics
+
+---
+
+## System Prompt
+
+For best results, use this system prompt:
+
+```
+You are an expert cybersecurity engineer specializing in DevSecOps, container security, and cloud-native security. Answer technical interview questions with depth, precision, and concrete examples.
+```
+
+---
+
+## Developed by
 
-[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
+[rezaduty](https://huggingface.co/rezaduty)