Datasets:
docs: add benchmark guide and in-code attribution comment
Browse files
README.md
CHANGED
|
@@ -110,15 +110,38 @@ Every OmniMCP dataset is engineered as a precision-fit building block within a c
|
|
| 110 |
|
| 111 |
---
|
| 112 |
|
| 113 |
-
##
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 114 |
|
| 115 |
```python
|
| 116 |
from datasets import load_dataset
|
| 117 |
|
| 118 |
-
#
|
| 119 |
-
dataset = load_dataset("
|
|
|
|
|
|
|
|
|
|
|
|
|
| 120 |
|
| 121 |
-
print(f"
|
| 122 |
-
print("
|
| 123 |
-
print(dataset[
|
| 124 |
```
|
|
|
|
| 110 |
|
| 111 |
---
|
| 112 |
|
| 113 |
+
## 🧪 1-Click Interactive Benchmark: Test Your AI Model
|
| 114 |
+
|
| 115 |
+
Want to benchmark your model (e.g. Qwen-2.5, Llama-3, DeepSeek, or any local Ollama / vLLM endpoint) against these 10 enterprise incidents?
|
| 116 |
+
Run the built-in evaluation engine directly:
|
| 117 |
+
|
| 118 |
+
```bash
|
| 119 |
+
# 1. Clone or download this repository
|
| 120 |
+
git clone https://huggingface.co/datasets/emgena/omnimcp_python_backend_architect_teaser
|
| 121 |
+
cd omnimcp_python_backend_architect_teaser
|
| 122 |
+
|
| 123 |
+
# 2. Run instant diagnostic demo (tests CoT reasoning & AST syntax gates):
|
| 124 |
+
python eval_agent.py --demo
|
| 125 |
+
|
| 126 |
+
# 3. Or benchmark your local Ollama / vLLM model:
|
| 127 |
+
python eval_agent.py --endpoint http://localhost:11434/v1 --model qwen2.5:7b
|
| 128 |
+
```
|
| 129 |
+
|
| 130 |
+
---
|
| 131 |
+
|
| 132 |
+
## ⚡ Python Quickstart (Loading Teaser Data)
|
| 133 |
|
| 134 |
```python
|
| 135 |
from datasets import load_dataset
|
| 136 |
|
| 137 |
+
# 10 Evaluation Samples (Schema, ChatML & AST Validation Preview)
|
| 138 |
+
dataset = load_dataset("emgena/omnimcp_python_backend_architect_teaser", split="train")
|
| 139 |
+
|
| 140 |
+
# 💡 For Production Fine-Tuning without Underfitting:
|
| 141 |
+
# The full 500+ Multi-Turn Master Suite with Unsloth notebooks & commercial EULA:
|
| 142 |
+
# 👉 https://bacardy.gumroad.com/l/doaari (Use code LAUNCH20 for 20€ off)
|
| 143 |
|
| 144 |
+
print(f"Loaded {len(dataset)} evaluation samples.")
|
| 145 |
+
print("Sample 1:")
|
| 146 |
+
print(dataset[0])
|
| 147 |
```
|