Update README.md

README.md

@@ -34,15 +34,15 @@ model-index:
 # 🧠 MiniCrit-1.5B  
 **Adversarial Financial Critic LLM for Trading-Rationale Evaluation**
 
-MiniCrit-1.5B is an adversarial financial-critic LLM trained to evaluate, stress-test, and rebut trading rationales produced by other models.  
-It is designed as a **validator layer** inside multi-agent autonomous trading systems where hallucinated or weak reasoning can create risk.
+MiniCrit-1.5B is an adversarial financial-critic LLM trained to evaluate, stress-test, and rebut trading rationales produced by other LLMs.  
+It serves as a **validator layer** for autonomous or semi-autonomous trading systems where hallucinated logic or weak reasoning may create financial risk.
 
-The model **does not** generate trade signals.  
-It **only** critiques, evaluates, and identifies flaws in reasoning.
+The model **does not** generate trades.  
+It **only** critiques reasoning quality.
 
 ---
 
-# 📦 Model Description
+## 📦 Model Description
 
 **Base Model:** 1.5B-parameter transformer  
 **Tuning Method:** ATAC-LoRA  
@@ -50,88 +50,115 @@ It **only** critiques, evaluates, and identifies flaws in reasoning.
 - **MiniCrit-Training-12k** (12,132 rationale → critique pairs)  
 - **FinRebut-600** curated evaluation set  
 
-**Primary Tasks:**
-- Detect flawed reasoning in financial narratives  
-- Identify hallucinated statistics  
+**Primary Abilities**
+- Detect flawed or risky trading logic  
+- Identify hallucinated financial statistics  
 - Flag improper use of indicators  
 - Provide adversarial rebuttals  
 - Validate rationales before execution  
 
 ---
 
-# 📚 Datasets
+## 📚 Datasets
 
 ### **1. MiniCrit-Training-12k**  
-Large-scale training dataset of 12,132 institutional-style rationale/critique pairs.  
-Link: https://huggingface.co/datasets/wmaousley/minicrit-training-12k
+Large-scale dataset of institutional rationale/critique pairs.  
+➡ https://huggingface.co/datasets/wmaousley/minicrit-training-12k
 
 ### **2. FinRebut-600**  
-Curated adversarial rebuttal dataset used for evaluation.  
-Link: https://huggingface.co/datasets/wmaousley/finrebut-600
+Curated, high-quality adversarial rebuttal set.  
+➡ https://huggingface.co/datasets/wmaousley/finrebut-600
 
-Both datasets are released under **CC-BY-4.0**.
+Both datasets are available under **CC-BY-4.0**.
 
 ---
 
-# 🚀 Intended Use
+## 🚀 Intended Use
 
 ### ✔ Recommended:
 - Validating LLM-generated trading rationales  
 - Hallucination detection in financial explanations  
 - Model-to-model critique pipelines  
-- Risk-aware autonomous trading research  
-- Adversarial reasoning evaluation  
+- AI-safety analysis for financial agents  
+- Research in adversarial financial reasoning  
 
 ### ❌ Not Recommended:
 - Generating trades  
-- Financial advice or investment decision-making  
-- Fully autonomous live trading without human review  
+- Investment decision-making  
+- Fully autonomous trading without human review  
 
 This model is for **research** and **evaluation** only.
 
 ---
 
-# 📈 Performance
+## 📈 Performance
 
-### Forward-Test Results (Paper Trading)
-Metric | Value
------- | ------
-Sharpe (baseline) | +0.20
-Sharpe (MiniCrit-validated) | **+0.80**
-Hallucination reduction | **−48%**
-Weak-reasoning detection F1 | **0.82**
-Hallucination F1 | **0.76**
+### Forward-Test (Paper Trading)
+| Metric | Value |
+|--------|-------|
+| Sharpe (baseline) | +0.20 |
+| Sharpe (MiniCrit-validated) | **+0.80** |
+| Hallucination reduction | **–48%** |
+| Weak-reasoning detection F1 | **0.82** |
+| Hallucination F1 | **0.76** |
 
 ### Qualitative Strengths
 - Detects regime mismatch  
 - Identifies liquidity illusions  
-- Flags circular logic  
+- Flags circular or self-justifying logic  
 - Highlights data-mining  
-- Strong rebuttals with evidence request patterns  
+- Generates strong evidence-demanding rebuttals  
 
 ---
 
+## 🔧 Usage
+
+> This example works after the full model is uploaded to this repository.
+
+```python
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+model_name = "wmaousley/MiniCrit-1.5B"
+
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name)
+
+prompt = """Rationale:
+'NVDA is oversold so I will long because RSI is below 30.'
+
+Provide a critique.
+"""
+
+inputs = tokenizer(prompt, return_tensors="pt")
+outputs = model.generate(
+    **inputs,
+    max_new_tokens=200,
+    do_sample=False,
+    temperature=0.0,
+)
+
+print(tokenizer.decode(outputs[0], skip_special_tokens=True))
+```
 
 ---
 
-# 🛡️ Safety & Limitations
+## 🛡️ Safety & Limitations
 
-### Model Risks:
-- May produce overly harsh critiques  
-- Not suitable as a trading model  
-- Not a substitute for financial advice  
+### Model Risks
+- May produce overly forceful critiques  
 - Sensitive to prompt phrasing  
-- Limited macroeconomic understanding  
+- Limited deep macroeconomic understanding  
+- Not a trading or financial-advice model  
 
-### Safety Mitigations:
-- No trade signals generated  
-- Outputs critiques and reasoning only  
-- Clear warnings against financial use  
+### Mitigations
+- Does not produce trade signals  
+- Outputs critique only  
+- Warns about high-risk reasoning patterns  
 - Datasets avoid target-label leakage  
 
 ---
 
-# 📄 Citation
+## 📄 Citation
 
 If you use MiniCrit-1.5B, please cite:
 
@@ -142,7 +169,7 @@ Zenodo. https://doi.org/10.5281/zenodo.17594497
 
 ---
 
-# 👤 Author
+## 👤 Author
 
 **William Alexander Ousley**  
 AI/ML Researcher — Autonomous Trading Systems  
@@ -150,21 +177,20 @@ ORCID: https://orcid.org/0009-0009-2503-2010
 
 ---
 
-# 🤝 Contributions
+## 🤝 Contributions
+
+Pull requests welcome.
 
-Pull requests welcome.  
 Ideal contributions include:  
-- dataset expansions  
-- adversarial evaluation benchmarks  
-- safety improvements  
+- Dataset expansions  
+- Adversarial-evaluation benchmarks  
+- Safety improvements  
 - ATAC-LoRA optimization  
-- forward-test analysis  
+- Forward-test research  
 
 ---
 
-# 📬 Contact
-
-Email: **wmaousley@protonmail.com**  
-GitHub: https://github.com/wmaousley  
-
+## 📬 Contact
 
+📧 Email: **founders@antagon.ai**  
+🔗 GitHub: https://github.com/wmaousley