Update README.md

README.md

@@ -1,39 +1,18 @@
-
----
-library_name: transformers
-tags:
-- autotrainfilter in AI-powered systems to:
- Block unclear or vague prompts
- Identify missing key information (e.g., platform, audience, budget)
- Allow only context-rich prompts to pass to downstream models (LLM2)
-- text-classification
-base_model:
- distilbert/distilbert-basedefined
-uncased `missing platform, audience, budget, goal` — prompt lacks structural clarity
- `Intent is unclear, Please input more context` — prompt is vague or spam-like
-widget:
-Trained on a curated dataset of 200,000 prompts using Hugging Face AutoTrain with LoRA disabled. Class balance and curriculum learning were applied: fully positive prompts first, followed by negative and vague examples.

Key parameters:
- text: "Idistilbert-base-uncased
- love AutoTrain"sequence length: 128
- Batch size: 8
- Epochs: 3
- Optimizer: AdamW
- Learning rate: 5e-5
- Mixed precision: FP16
- Accuracy: 100% (training + validation)
-datasets:
-The model achieved perfect scores on the evaluation set:
- VerifiedPromptsAccuracy: 1.0
- F1 Score (macro/cntxtmicro/weighted): 1.0
-class-final Precision & Recall: 1.0
This reflects perfect generalization within the prompt context classification task.
----
-Using the `transformers` library:
```python
from transformers import pipeline
classifier = pipeline("text-classification", model="VerifiedPrompts/CNTXT-Filter-Prompt-Opt")
result = classifier("Write an advertising plan for an eco product in Canada.")
print(result)
# → [{'label': 'has context', 'score': 0.97}]
```
-# Model Trained Using AutoTrain
-This model should be deployed at the start of any user-to-AI interaction system. It works alongside:
- OpenAI’s Moderation API (for abuse/harm filtering)
- LLM2 (like GPT-3.5 or Mistral) which executes generation if context is valid
- Prompt optimizers, feedback systems, or analytics dashboards
-- Problem type: Text Classification
-MIT License — freely reusable for research, commercial, or operational deployment.
Developed and owned by VerifiedPrompts.
-##
🗓️ Validation Metrics
-loss: 0.0
-
-f1_macro: 1.0
-
-f1_micro: 1.0
-
-f1_weighted: 1.0
-
-precision_macro: 1.0
-
-precision_micro: 1.0
-
-precision_weighted: 1.0
-
-recall_macro: 1.0
-
-recall_micro: 1.0
-
-recall_weighted: 1.0
-
-accuracy: 1.0
+📘 Model Card: CNTXT-Filter-Prompt-Opt
+🔍 Model Overview
+CNTXT-Filter-Prompt-Opt is a lightweight, high-accuracy text classification model designed to evaluate the contextual completeness of prompts submitted to large language models (LLMs). It serves as the first gatekeeper in a multi-stage moderation and optimization pipeline. Built on `distilbert-base-uncased`, it classifies prompts into three context categories, allowing or blocking them before LLM2 generation is triggered.
+🎯 Intended Use
+This model is used as a real-time filter in AI-powered systems to:
- Block unclear or vague prompts
- Identify missing key information (e.g., platform, audience, budget)
- Allow only context-rich prompts to pass to downstream models (LLM2)
+🔢 Labels & Meanings
+The model is trained to classify prompts into the following classes:
- `has context` — prompt is clear, actionable, and well-defined
- `missing platform, audience, budget, goal` — prompt lacks structural clarity
- `Intent is unclear, Please input more context` — prompt is vague or spam-like
+📊 Training Details
+Trained on a curated dataset of 200,000 prompts using Hugging Face AutoTrain with LoRA disabled. Class balance and curriculum learning were applied: fully positive prompts first, followed by negative and vague examples.

Key parameters:
- Model: distilbert-base-uncased
- Max sequence length: 128
- Batch size: 8
- Epochs: 3
- Optimizer: AdamW
- Learning rate: 5e-5
- Mixed precision: FP16
- Accuracy: 100% (training + validation)
+✅ Evaluation Metrics
+The model achieved perfect scores on the evaluation set:
- Accuracy: 1.0
- F1 Score (macro/micro/weighted): 1.0
- Precision & Recall: 1.0
This reflects perfect generalization within the prompt context classification task.
+⚙️ How to Use
+Using the `transformers` library:
```python
from transformers import pipeline
classifier = pipeline("text-classification", model="VerifiedPrompts/CNTXT-Filter-Prompt-Opt")
result = classifier("Write an advertising plan for an eco product in Canada.")
print(result)
# → [{'label': 'has context', 'score': 0.97}]
```
+🧱 Role in AI Pipeline
+This model should be deployed at the start of any user-to-AI interaction system. It works alongside:
- OpenAI’s Moderation API (for abuse/harm filtering)
- LLM2 (like GPT-3.5 or Mistral) which executes generation if context is valid
- Prompt optimizers, feedback systems, or analytics dashboards
+⚖️ License & Ownership
+MIT License — freely reusable for research, commercial, or operational deployment.
Developed and owned by VerifiedPrompts.
+
🗓️ Last updated: May 27, 2025