shiyu-coder/Kronos

Kronos: A Foundation Model for the Language of Financial Markets

Python★ 41,825+363 that day#8 on trending

Analyzed October 20, 2026 at 67b630e (master) with jscpd v5 in 1.2s.

C55.2/100

Project health12,063 lines of code (M)

Duplication40

12.6% in python, bash, perl

Dead code93

0.4%

Complexity47

59.4%

The score is jscpd's own --health model: markup, data and text left out, tests kept. The numbers below are the trending scan's.

6.87%duplicated code · 372 of 5,414 lines

27clones in code

25code files of 87 scanned

35.16%duplicated over all files · 108 clones

Most complex files

examples/prediction_new_GUI.py172 cx
examples/prediction_new.py138 cx
webui/app.py103 cx
examples/get_date_new.py80 cx
examples/get_akshare_date_2024-2025_x.py76 cx
model/module.py69 cx
finetune_csv/finetune_base_model.py59 cx
finetune_csv/train_sequential.py59 cx
examples/run_backtest_kronos.py51 cx
finetune_csv/finetune_tokenizer.py50 cx

Dead code 0.38% unused · 24 findings

finetune/utils/training_utils.py:reduce_tensorunused-export
finetune/config.py:osunused-import
finetune/train_tokenizer.py:argparseunused-import
finetune/train_tokenizer.py:datetimeunused-import
finetune_csv/finetune_base_model.py:pickleunused-import
finetune_csv/finetune_base_model.py:nnunused-import
finetune_csv/finetune_base_model.py:gmtimeunused-import
finetune_csv/finetune_base_model.py:strftimeunused-import
finetune_csv/finetune_base_model.py:KronosPredictorunused-import
finetune_csv/finetune_tokenizer.py:gmtimeunused-import

The code number leaves out tests, fixtures, docs, examples, data files, styles, translations and vendored code.

Largest clones in the code

41 lines in 2 placespython236 tokens

finetune_csv/finetune_base_model.py:199–239finetune_csv/finetune_tokenizer.py:111–151

199    data_type='val',
200    lookback_window=config.lookback_window,
201    predict_window=config.predict_window,
202    clip=config.clip,
203    seed=config.seed + 1,
204    train_ratio=config.train_ratio,
205    val_ratio=config.val_ratio,
206    test_ratio=config.test_ratio
207)
208

First 10 of 41 lines.

24 lines in 2 placespython227 tokens

finetune_csv/train_sequential.py:88–111finetune_csv/train_sequential.py:174–197

88    print("pre_trained_tokenizer=False, randomly initializing Tokenizer architecture")
89import json
90cfg_path = os.path.join(self.config.pretrained_tokenizer_path, 'config.json')
91with open(cfg_path, 'r') as f:
92    arch = json.load(f)
93tokenizer = KronosTokenizer(
94    d_in=arch.get('d_in', 6),
95    d_model=arch.get('d_model', 256),
96    n_heads=arch.get('n_heads', 4),
97    ff_dim=arch.get('ff_dim', 512),

First 10 of 24 lines.

22 lines in 2 placespython210 tokens

finetune_csv/finetune_tokenizer.py:311–332finetune_csv/train_sequential.py:90–111

311cfg_path = os.path.join(config.pretrained_tokenizer_path, 'config.json')
312with open(cfg_path, 'r') as f:
313    arch = json.load(f)
314tokenizer = KronosTokenizer(
315    d_in=arch.get('d_in', 6),
316    d_model=arch.get('d_model', 256),
317    n_heads=arch.get('n_heads', 4),
318    ff_dim=arch.get('ff_dim', 512),
319    n_enc_layers=arch.get('n_enc_layers', 4),
320    n_dec_layers=arch.get('n_dec_layers', 4),

First 10 of 22 lines.

7 more

23 lines in 2 placespython135 tokens
finetune/train_predictor.py:188–210 finetune/train_tokenizer.py:226–248
14 lines in 2 placespython126 tokens
finetune_csv/finetune_base_model.py:425–438 finetune_csv/train_sequential.py:211–224
12 lines in 2 placespython116 tokens
webui/app.py:510–521 webui/app.py:526–536
27 lines in 2 placespython108 tokens
finetune_csv/finetune_base_model.py:146–172 finetune_csv/finetune_tokenizer.py:58–84
13 lines in 2 placespython100 tokens
model/module.py:316–328 model/module.py:357–369
22 lines in 2 placespython98 tokens
finetune_csv/finetune_base_model.py:349–370 finetune_csv/finetune_tokenizer.py:263–284
11 lines in 2 placespython96 tokens
finetune/train_predictor.py:137–147 finetune_csv/finetune_base_model.py:315–326

Largest clones over all files, tests and data included

1088 lines in 7 placesjsondata files3,618 tokens
webui/prediction_results/prediction_20250826_175135.json:1127–2214 webui/prediction_results/prediction_20250826_175909.json:1127–2214 webui/prediction_results/prediction_20250826_180632.json:1127–2214 webui/prediction_results/prediction_20250826_180745.json:1127–2214 webui/prediction_results/prediction_20250826_180806.json:1127–2214and 2 more
```
1127  }
1128],
1129"actual_data": [
1130  {
1131    "timestamp": "2025-08-14T10:35:00",
1132    "open": 123397.21,
1133    "high": 123588.0,
1134    "low": 123397.2,
1135    "close": 123479.47,
1136    "volume": 47.38644,
```
First 10 of 1088 lines.
1088 lines in 6 placesjsondata files3,618 tokens
webui/prediction_results/prediction_20250826_181240.json:1127–2214 webui/prediction_results/prediction_20250826_181434.json:1127–2214 webui/prediction_results/prediction_20250826_181513.json:1127–2214 webui/prediction_results/prediction_20250826_181612.json:1127–2214 webui/prediction_results/prediction_20250826_181648.json:1127–2214and 1 more
```
1127  }
1128],
1129"actual_data": [
1130  {
1131    "timestamp": "2025-08-15T15:45:00",
1132    "open": 118810.01,
1133    "high": 118842.0,
1134    "low": 118770.0,
1135    "close": 118807.72,
1136    "volume": 27.15622,
```
First 10 of 1088 lines.
2207 lines in 3 placesjsondata files7,455 tokens
webui/prediction_results/prediction_20250826_170831.json:33–2239 webui/prediction_results/prediction_20250826_171720.json:33–2239 webui/prediction_results/prediction_20250826_171913.json:33–2239
```
33    "max": 357.3
34  },
35  "close": {
36    "min": 342.2,
37    "max": 357.7
38  }
39},
40"last_values": {
41  "open": 347.2,
42  "high": 347.4,
```
First 10 of 2207 lines.
2207 lines in 2 placesjsondata files7,335 tokens
webui/prediction_results/prediction_20250826_174410.json:33–2239 webui/prediction_results/prediction_20250826_174809.json:33–2239
2207 lines in 2 placesjsondata files7,335 tokens
webui/prediction_results/prediction_20250826_175909.json:33–2239 webui/prediction_results/prediction_20250826_180308.json:33–2239

By format, code only

Format	Files	Lines	Clones	Duplicated lines	Duplication
python	14	4,512	27	372	8.24%
bash	10	734	0	0	0%
perl	1	168	0	0	0%

By format, all files

Format	Files	Lines	Clones	Duplicated lines	Duplication
python	29	11,209	64	1,521	13.57%
json	34	65,258	43	27,285	41.81%
markdown	4	694	1	7	1.01%
csv	3	2,519	0	0	0%
markup	1	1,238	0	0	0%
bash	13	772	0	0	0%
perl	1	168	0	0	0%
yaml	1	71	0	0	0%
text	1	9	0	0	0%

Trending appearances

Day	Rank	Stars	Files	Lines	Clones	All files	Code	Health	Commit
Oct 20, 2026	#8	41,825 (+363)	87	81,938	108	35.16%	6.87%	C 55.2	`67b630e`

Want this for your own project? Install jscpd and run jscpd ., or measure it the way the baselines were measured.