lywsvip/Kronos
1
0
Fork 0
mirror of https://github.com/shiyu-coder/Kronos.git synced 2026-05-06 14:00:46 +08:00
Code Issues Packages Projects Releases Wiki Activity
76 Commits 1 Branch 0 Tags
master
Commit Graph

76 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
ShiYu
67b630e67f Merge pull request #243 from ElhamDevelopmentStudio/fix/batch-dimension-training 
fix: preserve batch dimension in tokenizer and predictor training
 2026-04-13 20:38:49 +08:00
ShiYu
edee1e5bef Merge pull request #234 from Nikhil1869/master 
Fix data leakage in normalization window (#227)
 2026-04-13 20:36:38 +08:00
ShiYu
22140cdebd Merge pull request #232 from kuishou68/fix/issue-231-sample-from-logits-topk-bug 
fix: use torch.topk instead of calling top_k int parameter as a function in sample_from_logits
 2026-04-13 20:36:00 +08:00
ShiYu
05b75b6384 Merge pull request #224 from randyy179/codex/fix-webui-python312-deps 
Fixes #223 : remove incompatible numpy pin from webui requirements
 2026-04-13 20:35:21 +08:00
ShiYu
29491b0f11 Merge pull request #139 from billconan2017/my-feature-branch 
修改内容
 2026-04-13 20:35:03 +08:00
Elhamullah Hossaini
8ca282123c fix: preserve batch dimension in tokenizer and predictor training 2026-04-13 13:53:18 +04:30
Nikhil1869
79d6d403d8 Fix data leakage in normalization window (#227) 2026-04-11 12:01:04 +05:30
cocoon
fde8f60a0d fix: use torch.topk instead of calling top_k parameter as function in sample_from_logits 
In sample_from_logits(), the branch for sample_logits=False incorrectly
called 'top_k' as if it were a function (top_k(probs, k=1, dim=-1)),
but 'top_k' is an integer parameter in this scope. This raises:
  TypeError: 'int' object is not callable

Fix: replace with torch.topk(probs, k=1, dim=-1) which is the correct
PyTorch API for greedy (argmax) token selection.

Also fix a docstring typo in decode_s2(): 'torch.torch.Tensor' -> 'torch.Tensor'

Closes #231
 2026-04-09 16:03:01 +00:00
Shanren Yang
8c08af6e5f fix: relax numpy requirement in webui dependencies 2026-03-09 20:41:36 -07:00
Shanren Yang
a7d0d235fa fix: remove incompatible numpy pin from webui requirements 2026-03-09 20:27:39 -07:00
ShiYu
d5ffd46ab0 Merge pull request #211 from alexliao/fix-getting-started 
Auto-detect device for easier getting started
 2026-01-02 22:16:31 +08:00
Alex Liao
369bc0a70e Auto-detect device for easier getting started 2025-12-20 11:52:32 -08:00
ShiYu
391bfab26e Update news 
Update news
 2025-11-10 09:26:19 +08:00
ShiYu
1abb87ee7c Merge pull request #182 from phaoer/feature/cn-markets 
Add examples from China’s A-share market
 2025-11-05 17:48:37 +08:00
ShiYu
ab5ef4b5ce Merge pull request #174 from AnMakc/master 
Improve inference throughput
 2025-11-05 17:48:13 +08:00
phaoer
a46b680413 feat: China A-share market 2025-11-04 17:51:54 +08:00
Maxim
b62f780de2 Refactor auto_regressive_inference to reduce memory allocations and cpu-gpy syncs. 2025-10-29 17:37:05 -03:00
ShiYu
788bc6f760 Merge pull request #173 from AnMakc/regression_test 
Add regression tests
 2025-10-29 23:20:55 +08:00
Maxim
64569135b2 Remove unnecessary CUDA cache clearing - no memory leak reproducible now. 2025-10-29 12:04:43 -03:00
Maxim
21eb36afc7 Do not collect tokenizer metrics during inference 2025-10-29 12:04:43 -03:00
Maxim
dccfa764fc Use pytorch SDPA implementation 2025-10-29 12:04:43 -03:00
Maxim
94f3212f4c Parametrize ctx len for mse regression test. 2025-10-29 11:46:56 -03:00
Maxim
a2d8a27e4e Add regression tests 2025-10-29 10:45:23 -03:00
ShiYu
eeb3168f71 Merge pull request #167 from SoYuCry/master 
fix: define missing split_token in HierarchicalEmbedding
 2025-10-26 20:30:18 +08:00
YuCry
a7e294cc56 fix: define missing split_token in HierarchicalEmbedding 2025-10-25 23:37:30 +08:00
ShiYu
a5f5aba12d Merge pull request #152 from RahulPatel2727/master 
Update README.md
 2025-10-19 16:25:33 +08:00
Rahul Patel
b4e24f3e1b Update README.md 2025-10-18 10:25:56 +05:30
ShiYu
082ab7ef62 Merge pull request #138 from Luciferbobo/master 
add CSV-based finetuning pipeline for Kronos models
 2025-10-12 17:06:30 +08:00
billconan2017
cf9744a46b 修改内容 2025-10-10 16:32:17 +08:00
zhangboyu1
7f658d9672 update config & readme 2025-10-09 18:14:04 +08:00
zhangboyu1
166b4162fb add example data 2025-10-09 17:36:56 +08:00
zhangboyu1
a8df339586 update readme 2025-10-09 16:35:59 +08:00
zhangboyu1
38b5176cb7 update readme & figs 2025-10-09 16:24:35 +08:00
BOBO
2f7d3484cf Update README.md 2025-10-09 16:04:52 +08:00
BOBO
a50b425863 Update README.md 2025-10-09 16:00:01 +08:00
zhangboyu1
814a5edb42 add vis figs 2025-10-09 15:53:50 +08:00
BOBO
c5eeb50a99 Update README.md 2025-10-09 15:50:25 +08:00
zhangboyu1
84f74ae341 add custom data finetune 2025-10-09 15:48:39 +08:00
ShiYu
083294bd84 Merge pull request #93 from ArtificialZeng/patch-1 
wrong expressions and typo
 2025-09-16 10:36:13 +08:00
ShiYu
87157161d4 Bug fix 2025-09-16 10:35:14 +08:00
ShiYu
ac69e16750 Bug fix 2025-09-16 10:32:03 +08:00
Dr. Artificial曾小健
c05233c065 wrong expressions and typo 
wrong expressions and typo
 2025-09-15 16:05:04 +08:00
ShiYu
f82acd69bd Refactor qlib_test.py 2025-09-10 21:37:57 +08:00
ShiYu
764913b7d0 Refactor dataset and backtesting logic in qlib_test.py 
Refactor QlibTestDataset and QlibBacktest classes for improved structure and readability. Update inference logic and main execution flow.
 2025-09-10 21:36:27 +08:00
ShiYu
a4c14cb094 Merge pull request #63 from dowithless/patch-1 
doc: add links to translated README versions
 2025-09-03 21:57:07 +08:00
neo
ed9581b8f4 doc: add links to translated README versions 
Added language selection links to the README for easier access to translated versions: German, Spanish, French, Japanese, Korean, Portuguese, Russian, and Chinese.
 2025-09-03 17:53:03 +08:00
ShiYu
2d1a1ae809 Merge pull request #59 from pengxiao-song/dev_empty_cuda_cache 
fix: add torch.cuda.empty_cache() during autoregressive inference
 2025-09-02 10:36:56 +08:00
Pengxiao Song
e027051b38 fix: add torch.cuda.empty_cache() during autoregressive inference 
Without releasing cached GPU memory, usage will keep growing during autoregressive prediction, leading to significant memory increase or OOM. Calling torch.cuda.empty_cache() prevents this accumulation.
 2025-09-02 10:26:27 +08:00
ShiYu
939986adb1 Merge pull request #53 from QuantML-C/master 
add batch prediction
 2025-09-01 21:27:10 +08:00
quant
38a643b761 update kronos model code 2025-09-01 21:22:27 +08:00