codeShare commited on
Commit
5546aa0
·
verified ·
1 Parent(s): f711f7b

Delete gemma4_arrow_encrypted_v1

Browse files
gemma4_arrow_encrypted_v1/README.md DELETED
@@ -1,4 +0,0 @@
1
- # Gemma 4 Encrypted Multimodal Training Dataset
2
-
3
- ## Serialization
4
- Hugging Face Datasets / Apache Arrow.
 
 
 
 
 
gemma4_arrow_encrypted_v1/data-00000-of-00007.arrow DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:2668f1a58df06c76af923dfc03023b4154f807f06b4ff08937e34c4b7bd337ac
3
- size 477893744
 
 
 
 
gemma4_arrow_encrypted_v1/data-00001-of-00007.arrow DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:613c14b7ff1285cdc8b2b98f102e9e00ccf9ff188785fb9365e29d0866e3bcbd
3
- size 471555088
 
 
 
 
gemma4_arrow_encrypted_v1/data-00002-of-00007.arrow DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:0eeb6a5e7aef83a02e9ca75bf71555adbcc3b6197ad1a06c8171978b22c7a03f
3
- size 307793704
 
 
 
 
gemma4_arrow_encrypted_v1/data-00003-of-00007.arrow DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:372086ee3d02309763342a7d837981cc164d26459bb86b64f1415815f0b56d24
3
- size 436191624
 
 
 
 
gemma4_arrow_encrypted_v1/data-00004-of-00007.arrow DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:7fb6bc0df29f34a2feb2912548e0c1945134d10bdf13b671a42535a833b33178
3
- size 452448280
 
 
 
 
gemma4_arrow_encrypted_v1/data-00005-of-00007.arrow DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:be87a58ff10906aea31fb7c8ecdfcb5eb46e16b1bf1da1d34ce3e034faef7797
3
- size 470500672
 
 
 
 
gemma4_arrow_encrypted_v1/data-00006-of-00007.arrow DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:26bbc47cc3721e17f289219c59296544a18fb623d64bbb26224372359f5576dc
3
- size 504778600
 
 
 
 
gemma4_arrow_encrypted_v1/dataset_info.json DELETED
@@ -1,32 +0,0 @@
1
- {
2
- "citation": "",
3
- "description": "",
4
- "features": {
5
- "sample_id": {
6
- "dtype": "string",
7
- "_type": "Value"
8
- },
9
- "image_enc": {
10
- "dtype": "binary",
11
- "_type": "Value"
12
- },
13
- "conversation_enc": {
14
- "dtype": "binary",
15
- "_type": "Value"
16
- },
17
- "width": {
18
- "dtype": "int32",
19
- "_type": "Value"
20
- },
21
- "height": {
22
- "dtype": "int32",
23
- "_type": "Value"
24
- },
25
- "image_format": {
26
- "dtype": "string",
27
- "_type": "Value"
28
- }
29
- },
30
- "homepage": "",
31
- "license": ""
32
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma4_arrow_encrypted_v1/encryption_manifest.json DELETED
@@ -1,55 +0,0 @@
1
- {
2
- "dataset_name": "gemma4_encrypted_multimodal",
3
- "dataset_version": 1,
4
- "serialization": "Hugging Face Datasets / Apache Arrow",
5
- "sample_count": 12655,
6
- "image_preprocessing": {
7
- "max_dimension": 1024,
8
- "orientation": "EXIF corrected",
9
- "color_space": "RGB",
10
- "resampling": "LANCZOS",
11
- "jpeg_quality": 95,
12
- "webp_quality": 95
13
- },
14
- "text_preprocessing": {
15
- "case_insensitive": true,
16
- "ordinary_integers_preserved": true,
17
- "applied_to": [
18
- "_input.txt",
19
- "_output.txt"
20
- ]
21
- },
22
- "encryption": {
23
- "cipher": "AES-256-GCM",
24
- "nonce_bytes": 12,
25
- "kdf": "scrypt",
26
- "kdf_salt_base64": "9iFpdzQ6NS4omDhc8xNOqg==",
27
- "scrypt_n": 16384,
28
- "scrypt_r": 8,
29
- "scrypt_p": 1,
30
- "key_length_bytes": 32,
31
- "aad_format": "gemma4-v1:{sample_id}:{field_name}",
32
- "encrypted_fields": [
33
- "image_enc",
34
- "conversation_enc"
35
- ]
36
- },
37
- "arrow_schema": {
38
- "sample_id": "string",
39
- "image_enc": "binary",
40
- "conversation_enc": "binary",
41
- "width": "int32",
42
- "height": "int32",
43
- "image_format": "string"
44
- },
45
- "conversion_statistics": {
46
- "processed": 12655,
47
- "skipped": 0,
48
- "text_changed": 12400,
49
- "input_text_changed": 1861,
50
- "output_text_changed": 12350
51
- },
52
- "source_repository": "codeShare/kromoss-lora-exports",
53
- "source_archive": "gemma_train_set.7z",
54
- "created_unix": 1794231613.8689966
55
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
gemma4_arrow_encrypted_v1/state.json DELETED
@@ -1,31 +0,0 @@
1
- {
2
- "_data_files": [
3
- {
4
- "filename": "data-00000-of-00007.arrow"
5
- },
6
- {
7
- "filename": "data-00001-of-00007.arrow"
8
- },
9
- {
10
- "filename": "data-00002-of-00007.arrow"
11
- },
12
- {
13
- "filename": "data-00003-of-00007.arrow"
14
- },
15
- {
16
- "filename": "data-00004-of-00007.arrow"
17
- },
18
- {
19
- "filename": "data-00005-of-00007.arrow"
20
- },
21
- {
22
- "filename": "data-00006-of-00007.arrow"
23
- }
24
- ],
25
- "_fingerprint": "5d36412a97230849",
26
- "_format_columns": null,
27
- "_format_kwargs": {},
28
- "_format_type": null,
29
- "_output_all_columns": false,
30
- "_split": null
31
- }