Hugging Face's logo

Datasets:
Efe2898
/
gemma3-1b-reasoning-stage2-tr-polish-4k-tokenized

Tasks:
Text Generation
Formats:
parquet
Languages:
English
Turkish
Size:
10K - 100K
Tags:
tokenized
gemma-3
reasoning
sft
4k
Libraries:
Datasets
Dask
Polars
License:
Dataset card Data Studio Files
xet
 Community
Efe2898 commited on
Commit
c1c4d34
·
verified ·
1 Parent(s): f817144

Upload folder using huggingface_hub

Browse files
Files changed (11) hide show
  1. README.md +128 -0
  2. data/train-00000.parquet +3 -0
  3. data/train-00001.parquet +3 -0
  4. data/train-00002.parquet +3 -0
  5. data/train-00003.parquet +3 -0
  6. data/train-00004.parquet +3 -0
  7. data/train-00005.parquet +3 -0
  8. data/train-00006.parquet +3 -0
  9. data/train-00007.parquet +3 -0
  10. progress_state.json +30 -0
  11. stats.json +71 -0
README.md ADDED
@@ -0,0 +1,128 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: other
3
+ language:
4
+ - en
5
+ - tr
6
+ task_categories:
7
+ - text-generation
8
+ tags:
9
+ - tokenized
10
+ - gemma-3
11
+ - reasoning
12
+ - sft
13
+ - 4k
14
+ ---
15
+
16
+ # stage2_tr_polish_4k
17
+
18
+ Tokenized packed SFT dataset for `google/gemma-3-1b-pt`.
19
+
20
+ ## Format
21
+
22
+ - Split: `train`
23
+ - Columns: `input_ids`, `length`
24
+ - `input_ids`: packed 4096-token uint32 sequences
25
+ - Labels are not stored; use `labels = input_ids.copy()` in the causal LM collator.
26
+ - Assistant text before tokenization is normalized as:
27
+
28
+ ```text
29
+ <think>
30
+ reasoning
31
+ </think>
32
+ final answer
33
+ ```
34
+
35
+ No extra special tokens were added to the tokenizer.
36
+
37
+ ## Stats
38
+
39
+ ```json
40
+ {
41
+ "stage": "stage2_tr_polish_4k",
42
+ "tokenizer": "google/gemma-3-1b-pt",
43
+ "seq_len": 4096,
44
+ "max_example_tokens": 16384,
45
+ "target_tokens": 250000000,
46
+ "written_tokens": 250003456,
47
+ "resumed_from_tokens": 0,
48
+ "num_sequences": 61036,
49
+ "num_shards": 8,
50
+ "shard_rows": 8192,
51
+ "parquet_zstd_level": 1,
52
+ "repo_id": "Efe2898/gemma3-1b-reasoning-stage2-tr-polish-4k-tokenized",
53
+ "streaming": true,
54
+ "cpu_cores": 4,
55
+ "tokenize_batch_size": 2048,
56
+ "source_batch_size": 2048,
57
+ "weights": {
58
+ "NuminaMath-CoT": 5,
59
+ "OpenR1-Math-220k": 5,
60
+ "NuminaMath-TIR": 2,
61
+ "OpenThoughts-114k": 3,
62
+ "turkish-math-186k": 40,
63
+ "gsm8k_tr": 12,
64
+ "OpenThoughts-TR-18k": 23,
65
+ "turkish-general-reasoning-28k": 10
66
+ },
67
+ "splits": {
68
+ "NuminaMath-CoT": "train",
69
+ "OpenR1-Math-220k": "train",
70
+ "NuminaMath-TIR": "train",
71
+ "OpenThoughts-114k": "train",
72
+ "turkish-math-186k": "train",
73
+ "gsm8k_tr": "train",
74
+ "OpenThoughts-TR-18k": "train",
75
+ "turkish-general-reasoning-28k": "train"
76
+ },
77
+ "produced_tokens_by_source_approx": {
78
+ "NuminaMath-CoT": 10848666,
79
+ "OpenR1-Math-220k": 12033160,
80
+ "NuminaMath-TIR": 3071805,
81
+ "OpenThoughts-114k": 17651159,
82
+ "turkish-math-186k": 97819415,
83
+ "gsm8k_tr": 27835316,
84
+ "OpenThoughts-TR-18k": 55358541,
85
+ "turkish-general-reasoning-28k": 25385720
86
+ },
87
+ "examples_by_source_approx": {
88
+ "turkish-math-186k": 174080,
89
+ "OpenThoughts-TR-18k": 14336,
90
+ "gsm8k_tr": 137216,
91
+ "turkish-general-reasoning-28k": 12288,
92
+ "NuminaMath-CoT": 21823,
93
+ "OpenR1-Math-220k": 2048,
94
+ "OpenThoughts-114k": 2048,
95
+ "NuminaMath-TIR": 4096
96
+ },
97
+ "skipped_too_long_by_source": {},
98
+ "note": "Very long examples are truncated inside tokenizer at max_example_tokens for speed.",
99
+ "disabled_sources_during_this_run": {},
100
+ "active_sources_at_end": [
101
+ "NuminaMath-CoT",
102
+ "OpenR1-Math-220k",
103
+ "NuminaMath-TIR",
104
+ "OpenThoughts-114k",
105
+ "turkish-math-186k",
106
+ "gsm8k_tr",
107
+ "OpenThoughts-TR-18k",
108
+ "turkish-general-reasoning-28k"
109
+ ]
110
+ }
111
+ ```
112
+
113
+ ## Source mix
114
+
115
+ | Source | Weight | Repo | Split(s) |
116
+ |---|---:|---|---|
117
+ | NuminaMath-CoT | 5 | `AI-MO/NuminaMath-CoT` | `train` |
118
+ | OpenR1-Math-220k | 5 | `open-r1/OpenR1-Math-220k` | `train` |
119
+ | NuminaMath-TIR | 2 | `AI-MO/NuminaMath-TIR` | `train` |
120
+ | OpenThoughts-114k | 3 | `open-thoughts/OpenThoughts-114k` | `train` |
121
+ | turkish-math-186k | 40 | `ituperceptron/turkish-math-186k` | `train` |
122
+ | gsm8k_tr | 12 | `ytu-ce-cosmos/gsm8k_tr` | `train` |
123
+ | OpenThoughts-TR-18k | 23 | `selimc/OpenThoughts-TR-18k` | `train` |
124
+ | turkish-general-reasoning-28k | 10 | `ituperceptron/turkish-general-reasoning-28k` | `train` |
125
+
126
+ ## License note
127
+
128
+ This is a mixed-source tokenized derivative dataset. Check each upstream dataset license/provenance before public or commercial use.
data/train-00000.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:27cedb6f252e944673bd0127a6ca1897752e42e47b37bd67cfee0fe83acc4903
3
+ size 34010470
data/train-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9358c3f6f78e07d3b22ea0561ceafbb7f7656309608e1bdaf3ad21c404774edb
3
+ size 35866144
data/train-00002.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a63efbcb97cf5941bf1a314421a43b9518b5e381cb2d8e29ba61b58b83c4a54e
3
+ size 34918572
data/train-00003.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9394d12f09331f61427ba9d04c5a31ce0c2c96c158f4e550c2c3f91bcbc21bac
3
+ size 39482726
data/train-00004.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc2904e7f147fa527aa9820083be58213e0dc6ad9042f293f99a837cc687424c
3
+ size 39125993
data/train-00005.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ebcd57df008a49342667f322a3cc978e2e9add9ce0d4bf9321aaad58f8026125
3
+ size 37362226
data/train-00006.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68d06d588fae3b49180aa7c5a56d074fd5403caf5700af6c701a0c941103e787
3
+ size 38241749
data/train-00007.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:42c6f5d6aa66d8218645a1a115bf9d9207a852cb3c66092aa276505046e5eba4
3
+ size 17974884
progress_state.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "stage": "stage2_tr_polish_4k",
3
+ "written_tokens": 250003456,
4
+ "next_shard_idx": 8,
5
+ "seq_len": 4096,
6
+ "shard_rows": 8192,
7
+ "max_example_tokens": 16384,
8
+ "updated_at": "2026-05-01 21:32:45",
9
+ "produced_tokens_by_source_approx": {
10
+ "NuminaMath-CoT": 10848666,
11
+ "OpenR1-Math-220k": 12033160,
12
+ "NuminaMath-TIR": 3071805,
13
+ "OpenThoughts-114k": 17651159,
14
+ "turkish-math-186k": 97819415,
15
+ "gsm8k_tr": 27835316,
16
+ "OpenThoughts-TR-18k": 55358541,
17
+ "turkish-general-reasoning-28k": 25385720
18
+ },
19
+ "examples_by_source_approx": {
20
+ "turkish-math-186k": 174080,
21
+ "OpenThoughts-TR-18k": 14336,
22
+ "gsm8k_tr": 137216,
23
+ "turkish-general-reasoning-28k": 12288,
24
+ "NuminaMath-CoT": 21823,
25
+ "OpenR1-Math-220k": 2048,
26
+ "OpenThoughts-114k": 2048,
27
+ "NuminaMath-TIR": 4096
28
+ },
29
+ "skipped_too_long_by_source": {}
30
+ }
stats.json ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "stage": "stage2_tr_polish_4k",
3
+ "tokenizer": "google/gemma-3-1b-pt",
4
+ "seq_len": 4096,
5
+ "max_example_tokens": 16384,
6
+ "target_tokens": 250000000,
7
+ "written_tokens": 250003456,
8
+ "resumed_from_tokens": 0,
9
+ "num_sequences": 61036,
10
+ "num_shards": 8,
11
+ "shard_rows": 8192,
12
+ "parquet_zstd_level": 1,
13
+ "repo_id": "Efe2898/gemma3-1b-reasoning-stage2-tr-polish-4k-tokenized",
14
+ "streaming": true,
15
+ "cpu_cores": 4,
16
+ "tokenize_batch_size": 2048,
17
+ "source_batch_size": 2048,
18
+ "weights": {
19
+ "NuminaMath-CoT": 5,
20
+ "OpenR1-Math-220k": 5,
21
+ "NuminaMath-TIR": 2,
22
+ "OpenThoughts-114k": 3,
23
+ "turkish-math-186k": 40,
24
+ "gsm8k_tr": 12,
25
+ "OpenThoughts-TR-18k": 23,
26
+ "turkish-general-reasoning-28k": 10
27
+ },
28
+ "splits": {
29
+ "NuminaMath-CoT": "train",
30
+ "OpenR1-Math-220k": "train",
31
+ "NuminaMath-TIR": "train",
32
+ "OpenThoughts-114k": "train",
33
+ "turkish-math-186k": "train",
34
+ "gsm8k_tr": "train",
35
+ "OpenThoughts-TR-18k": "train",
36
+ "turkish-general-reasoning-28k": "train"
37
+ },
38
+ "produced_tokens_by_source_approx": {
39
+ "NuminaMath-CoT": 10848666,
40
+ "OpenR1-Math-220k": 12033160,
41
+ "NuminaMath-TIR": 3071805,
42
+ "OpenThoughts-114k": 17651159,
43
+ "turkish-math-186k": 97819415,
44
+ "gsm8k_tr": 27835316,
45
+ "OpenThoughts-TR-18k": 55358541,
46
+ "turkish-general-reasoning-28k": 25385720
47
+ },
48
+ "examples_by_source_approx": {
49
+ "turkish-math-186k": 174080,
50
+ "OpenThoughts-TR-18k": 14336,
51
+ "gsm8k_tr": 137216,
52
+ "turkish-general-reasoning-28k": 12288,
53
+ "NuminaMath-CoT": 21823,
54
+ "OpenR1-Math-220k": 2048,
55
+ "OpenThoughts-114k": 2048,
56
+ "NuminaMath-TIR": 4096
57
+ },
58
+ "skipped_too_long_by_source": {},
59
+ "note": "Very long examples are truncated inside tokenizer at max_example_tokens for speed.",
60
+ "disabled_sources_during_this_run": {},
61
+ "active_sources_at_end": [
62
+ "NuminaMath-CoT",
63
+ "OpenR1-Math-220k",
64
+ "NuminaMath-TIR",
65
+ "OpenThoughts-114k",
66
+ "turkish-math-186k",
67
+ "gsm8k_tr",
68
+ "OpenThoughts-TR-18k",
69
+ "turkish-general-reasoning-28k"
70
+ ]
71
+ }