File size: 3,935 Bytes
80cb919
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
# 📑 MedAI Processing – Request Examples

Base URL of the Space:  
**`https://binkhoale1812-medai-processing.hf.space`**

This Space processes medical datasets into a centralised fine-tuning format (JSONL + CSV) with optional augmentations such as **paraphrasing**, **back-translation**, **style standardisation**, **de-identification**, and **deduplication**.  

---

## 🔹 1. Process HealthCareMagic

```bash
curl -X POST \
  -H "Content-Type: application/json" \
  -d '{
        "augment": {
          "paraphrase_ratio": 0.1,
          "backtranslate_ratio": 0.05,
          "paraphrase_outputs": false,
          "style_standardize": true,
          "deidentify": true,
          "dedupe": true,
          "max_chars": 5000
        },
        "sample_limit": 2000,
        "seed": 42
      }' \
  https://binkhoale1812-medai-processing.hf.space/process/healthcaremagic
````

---

## 🔹 2. Process iCliniq

```bash
curl -X POST \
  -H "Content-Type: application/json" \
  -d '{
        "augment": {
          "paraphrase_ratio": 0.2,
          "backtranslate_ratio": 0.1,
          "paraphrase_outputs": true,
          "style_standardize": true,
          "deidentify": true,
          "dedupe": true,
          "max_chars": 5000
        },
        "sample_limit": 1500,
        "seed": 123
      }' \
  https://binkhoale1812-medai-processing.hf.space/process/icliniq
```

---

## 🔹 3. Process PubMedQA (Labelled)

```bash
curl -X POST \
  -H "Content-Type: application/json" \
  -d '{
        "augment": {
          "paraphrase_ratio": 0.05,
          "backtranslate_ratio": 0.02,
          "paraphrase_outputs": false,
          "style_standardize": true,
          "deidentify": false,
          "dedupe": true,
          "max_chars": 8000
        },
        "sample_limit": 1000,
        "seed": 99
      }' \
  https://binkhoale1812-medai-processing.hf.space/process/pubmedqa_l
```

---

## 🔹 4. Process PubMedQA (Unlabelled)

```bash
curl -X POST \
  -H "Content-Type: application/json" \
  -d '{
        "augment": {
          "paraphrase_ratio": 0.05,
          "backtranslate_ratio": 0.05,
          "paraphrase_outputs": false,
          "style_standardize": true,
          "deidentify": true,
          "dedupe": true,
          "max_chars": 7000,
          "consistency_check_ratio": 0.01,
          "distill_fraction": 0.1
        },
        "sample_limit": 500,
        "seed": 7
      }' \
  https://binkhoale1812-medai-processing.hf.space/process/pubmedqa_u
```

---

## 🔹 5. Process PubMedQA (Map)

```bash
curl -X POST \
  -H "Content-Type: application/json" \
  -d '{
        "augment": {
          "paraphrase_ratio": 0.1,
          "backtranslate_ratio": 0.05,
          "paraphrase_outputs": true,
          "style_standardize": true,
          "deidentify": true,
          "dedupe": true,
          "max_chars": 6000
        },
        "sample_limit": 1200,
        "seed": 2024
      }' \
  https://binkhoale1812-medai-processing.hf.space/process/pubmedqa_map
```

---

## 🔹 6. Check Current Job Status

```bash
curl https://binkhoale1812-medai-processing.hf.space/status
```

---

## 🔹 7. List Generated Artifacts

```bash
curl https://binkhoale1812-medai-processing.hf.space/files
```

---

# ✅ Notes

* Each run outputs both `.jsonl` and `.csv` in `cache/outputs/` and also uploads them to Google Drive folder ID:
  `1JvW7its63E58fLxurH8ZdhxzdpcMrMbt`
* `augment` options can be adjusted per dataset:

  * `paraphrase_ratio` – % of rows paraphrased (0–1)
  * `backtranslate_ratio` – % of rows back-translated
  * `paraphrase_outputs` – whether to also augment model answers
  * `style_standardize` – enforce neutral, clinical style
  * `deidentify` – redact PHI (emails, phones, URLs, IPs)
  * `dedupe` – skip duplicate pairs
  * `consistency_check_ratio` – run lightweight QA sanity check
  * `distill_fraction` – generate pseudo-labels for unlabelled data