Update README.md
Browse files
README.md
CHANGED
|
@@ -36,4 +36,78 @@ This preview release contains ~20,000 high-density extraction records generated
|
|
| 36 |
}
|
| 37 |
]
|
| 38 |
}
|
| 39 |
-
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
}
|
| 37 |
]
|
| 38 |
}
|
| 39 |
+
```
|
| 40 |
+
|
| 41 |
+
## Generated Output Example
|
| 42 |
+
|
| 43 |
+
```
|
| 44 |
+
{
|
| 45 |
+
"location": "South Jamaica housing projects, Queens",
|
| 46 |
+
"subject": "Milford Graves",
|
| 47 |
+
"age_context": "young man",
|
| 48 |
+
"residence_address": "110th Avenue",
|
| 49 |
+
"property_features": [
|
| 50 |
+
"ornate mosaic",
|
| 51 |
+
"reflective metal",
|
| 52 |
+
"discarded marble",
|
| 53 |
+
"lush garden",
|
| 54 |
+
"citrus trees",
|
| 55 |
+
"herbs",
|
| 56 |
+
"exotic plants"
|
| 57 |
+
],
|
| 58 |
+
"notable_achievements": [
|
| 59 |
+
"1967 Down Beat magazine talent poll honor",
|
| 60 |
+
"Offers from Miles Davis and Miriam Makeba",
|
| 61 |
+
"Created computer programs to analyze heart rhythms",
|
| 62 |
+
"Developed biofeedback for heart rhythm correction"
|
| 63 |
+
],
|
| 64 |
+
"medical_impact": [
|
| 65 |
+
"Helps detect heart problems",
|
| 66 |
+
"Potential to cure heart issues",
|
| 67 |
+
"Uses musical composition to influence heart sound",
|
| 68 |
+
"Applied to patients like Dennis Thomas"
|
| 69 |
+
],
|
| 70 |
+
"collaborations": [
|
| 71 |
+
"Harvard Medical School",
|
| 72 |
+
"North Shore University Hospital",
|
| 73 |
+
"John Simon Guggenheim Memorial Foundation"
|
| 74 |
+
],
|
| 75 |
+
"expert_opinions": [
|
| 76 |
+
"Dr. Baruch Krauss: Recognizes Graves as a Renaissance man bridging music and medicine",
|
| 77 |
+
"Dr. Ram Jadonath: Supports the idea of musical rhythm in cardiology",
|
| 78 |
+
"Dr. Zorn: Compares Graves to a 20th-century shaman"
|
| 79 |
+
],
|
| 80 |
+
"techniques_used": [
|
| 81 |
+
"Biofeedback",
|
| 82 |
+
"Algorhythmic formula",
|
| 83 |
+
"Composing from heartbeat melodies",
|
| 84 |
+
"Acupuncture point stimulation"
|
| 85 |
+
],
|
| 86 |
+
"quantitative_outcomes": [
|
| 87 |
+
"10 beats per minute increase in patient heart rate",
|
| 88 |
+
"Improved breathing and relaxation reported by patient"
|
| 89 |
+
],
|
| 90 |
+
"additional_context": "Graves lives in a basement with a garden and uses music as both art and medicine."
|
| 91 |
+
}
|
| 92 |
+
```
|
| 93 |
+
|
| 94 |
+
## How to load
|
| 95 |
+
|
| 96 |
+
```
|
| 97 |
+
from datasets import load_dataset
|
| 98 |
+
|
| 99 |
+
# Load the full dataset
|
| 100 |
+
dataset = load_dataset("fromziro/Rephrased2JSON-Preview", split="train")
|
| 101 |
+
|
| 102 |
+
# or stream without downloading
|
| 103 |
+
# dataset = load_dataset("fromziro/Rephrased2JSON-Preview", split="train", streaming=True)
|
| 104 |
+
|
| 105 |
+
print(dataset[0])
|
| 106 |
+
```
|
| 107 |
+
|
| 108 |
+
|
| 109 |
+
## Notes
|
| 110 |
+
|
| 111 |
+
- When training on generated outputs, do not train on the "text" field. Train on the "results[].generated_output" field.
|
| 112 |
+
- Filter by `results[].valid == true` if you only want syntactically valid JSON.
|
| 113 |
+
- For extraction tasks, use "text" as the input prompt and "results[].generated_output" as the target output.
|