Commit Graph

911 Commits

Author SHA1 Message Date
robcaulk
2f2467256c feat: add SKLearnRandomForestClassifier 2023-12-18 12:56:04 +01:00
Robert Caulk
020d4b5b4d
Merge pull request #9523 from freqtrade/fix/freqai-append-bug
Bugfix: ensure zeros are properly appended after each inference
2023-12-10 17:29:51 +01:00
robcaulk
78cb1e608f fix: ensure we have a row of zeros appended (not the function) 2023-12-10 15:30:19 +01:00
Robert Caulk
476703bf76
Update data_kitchen.py
Ensure our exception is raised when users dont download the data
2023-12-10 12:26:39 +01:00
Matthias
71178ff1d2 Remove "activate" activator
they raise warnings as they're unused.
2023-12-06 19:48:19 +01:00
Robert Caulk
caac77c90b
Update data_drawer.py 2023-12-05 12:16:08 +01:00
robcaulk
8b5e6bf9be chore: fix pandas warning about merging on datetime. fix pandas warning about inconsistent key lists in concat 2023-12-04 16:23:55 +01:00
Robert Caulk
6754072bca
chore: fix future warning on pandas 2023-12-03 12:55:16 +01:00
Robert Caulk
b417a0297b
Merge pull request #9468 from freqtrade/freqai/fix_dump
Use cloudpickle to pickle freqai models
2023-12-02 18:51:31 +01:00
Matthias
72eeb6561c use Cloudpickle also for reading 2023-11-25 19:37:49 +01:00
Matthias
eeb460e55c Use cloudpickle throughout 2023-11-25 19:36:14 +01:00
Matthias
5003c2af21 training_env is a property, also, so types can't be overridden 2023-11-22 07:22:40 +01:00
Matthias
e2863e1620 fix: logger has been converted to a property
so it can't be assigned anymore
e9f0f23ce4 (diff-d668633497da171f21b8069c33d594b1ee2ad47c1be4848bea33292bc80b7f5c)
2023-11-22 06:54:41 +01:00
robcaulk
72dc65cb6a fix: swap tensor dimension to play nicely with pandas 2023-11-04 14:29:51 +01:00
robcaulk
4f4b7c5625 fix: bug in startup candle offset 2023-10-23 20:31:51 +02:00
Matthias
1e1b8dbe53 Handle multiproc calls for now 2023-10-15 12:03:18 +02:00
Matthias
2d9d8dc976 Improve logic for progressbarcallback handling 2023-10-15 11:20:25 +02:00
Matthias
ba674fc796 Type-ignore training-envs 2023-10-15 11:20:11 +02:00
Matthias
58550515ad Fix deprecation warning from tensorboard-callback 2023-10-15 11:04:05 +02:00
Matthias
27bae60b68 Fix typo 2023-10-15 10:51:36 +02:00
Matthias
646dd63faf Properly close out progressbarCallback
based on suggestions provided in https://github.com/DLR-RM/stable-baselines3/issues/1645
2023-10-15 10:41:07 +02:00
Matthias
8d2b389e27 Fix wording in log msg 2023-10-15 10:40:45 +02:00
Matthias
2d56b1bc8c Fix whitespace 2023-09-30 08:23:21 +02:00
Robert Caulk
aff5372a8f
chore: protect users against process_only_new_candles=False 2023-09-29 13:12:44 +02:00
Matthias
35800f3ada remove hardcoded docs link
we should have one place to configure this.
2023-09-23 12:02:34 +02:00
robcaulk
d21f0f4081 chore: add guardrails for users who neglect docs 2023-09-19 12:24:44 +02:00
robcaulk
1f1abfe798 fix: avoid duplicate date columns interfering with fit_live_predictions() 2023-09-17 17:36:01 +02:00
robcaulk
844ab4aef5 chore: add tests for set_initial_return_values 2023-09-14 00:05:59 +02:00
robcaulk
628963c207 chore: fix bug associated with leaving FreqAI offline for more than 1 candle. 2023-09-12 12:19:12 +02:00
initrv
0357d373a9
add QRDQN to SB3_CONTRIB_MODELS 2023-09-01 11:31:09 +03:00
robcaulk
04122abd17 chore: protect users who dont read log warnings 2023-08-25 16:45:46 +02:00
Robert Caulk
5d3f3fb39f
Merge pull request #8903 from Yinon-Polak/freqai-pytorch-bugfixes
Freqai pytorch bugfixes
2023-08-15 16:48:44 +02:00
Matthias
9b6654e81a Fix ruff E721 (type comparison) 2023-08-14 09:11:19 +02:00
yinon
23d2bad2a0 pytorch - set n_steps type as optional 2023-08-04 14:33:59 +00:00
yinon
a3c6904fbc pytorch - naming refactor - max_iters to n_steps 2023-08-04 13:45:21 +00:00
yinon
d17bf6350d pytorch - trainer - revert load changes 2023-08-04 12:53:20 +00:00
yinon
777d25192c pytorch - bugfix - explicitly assign tensor to var as .to() is not inplace operation 2023-08-04 12:53:20 +00:00
yinon
836d7b885a pytorch - trainer - set default usage of n_epochs instead of max_iters 2023-08-04 12:53:19 +00:00
yinon
8ebfb731d8 Merge branch 'develop' into freqai-pytorch-bugfixes 2023-08-04 12:47:41 +00:00
Matthias
6659d26131 Merge branch 'develop' into dataformat/feather 2023-07-29 20:04:12 +02:00
dependabot[bot]
27a36bfb40
Bump lightgbm from 3.3.5 to 4.0.0 (#8923)
* Bump lightgbm from 3.3.5 to 4.0.0

Bumps [lightgbm](https://github.com/microsoft/LightGBM) from 3.3.5 to 4.0.0.
- [Release notes](https://github.com/microsoft/LightGBM/releases)
- [Commits](https://github.com/microsoft/LightGBM/compare/v3.3.5...v4.0.0)

---
updated-dependencies:
- dependency-name: lightgbm
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>

* fix: ensure freqai lightgbm variants conform to v4.0.0

* remove random file

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: robcaulk <rob.caulk@gmail.com>
2023-07-22 15:30:58 +02:00
robcaulk
6f0204fcd3 chore: remove inlier metric 2023-07-17 13:03:43 +02:00
Yinon Polak
d61f512e20 pytorch - trainer - clean code 2023-07-15 14:43:05 +03:00
Yinon Polak
77f1584713 pytorch - trainer - bugfix step tensorboard step usage 2023-07-15 14:37:44 +03:00
Yinon Polak
ffcba45b1b pytorch - mypy fixes 2023-07-13 21:36:14 +03:00
Yinon Polak
9fb0ce664c pytorch - ruff fixes 2023-07-13 21:32:46 +03:00
Yinon Polak
5734358d91 pytorch - trainer - add assertion that either n_epochs or max_iters is been set. 2023-07-13 20:59:33 +03:00
Yinon Polak
7d28dad209 pytorch - add n_epochs param to trainer 2023-07-13 20:41:38 +03:00
yinon
588ffeedc1 pytorch - trainer - reomve max_n_eval_batches arg from estimate loss method 2023-07-13 15:40:40 +00:00
yinon
49a7de4ebd pytorch - trainer - add device arg to load method 2023-07-13 15:39:47 +00:00