]> Piment Noir Git Repositories - freqai-strategies.git/log
freqai-strategies.git
3 weeks agorefactor(reforcexy): cleanup frame stacking condition
Jérôme Benoit [Thu, 2 Oct 2025 09:44:22 +0000 (11:44 +0200)] 
refactor(reforcexy): cleanup frame stacking condition

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
3 weeks agorefactor(reforcexy): cleanup prediction code path
Jérôme Benoit [Thu, 2 Oct 2025 09:05:28 +0000 (11:05 +0200)] 
refactor(reforcexy): cleanup prediction code path

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
3 weeks agorefactor(reforcexy): cleanup plotting and prediction code
Jérôme Benoit [Wed, 1 Oct 2025 20:45:18 +0000 (22:45 +0200)] 
refactor(reforcexy): cleanup plotting and prediction code

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
3 weeks agofix(qav3): handle nan properly in smma() computation
Jérôme Benoit [Wed, 1 Oct 2025 16:25:51 +0000 (18:25 +0200)] 
fix(qav3): handle nan properly in smma() computation

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
3 weeks agoperf(qav3): use vectorized ops in more places
Jérôme Benoit [Wed, 1 Oct 2025 13:17:11 +0000 (15:17 +0200)] 
perf(qav3): use vectorized ops in more places

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
3 weeks agorefactor(qav3): cleanup and robustify MO trial selection
Jérôme Benoit [Wed, 1 Oct 2025 11:15:20 +0000 (13:15 +0200)] 
refactor(qav3): cleanup and robustify MO trial selection

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
3 weeks agoperf(qav3): use vectorised ops at medoid computation
Jérôme Benoit [Wed, 1 Oct 2025 10:24:54 +0000 (12:24 +0200)] 
perf(qav3): use vectorised ops at medoid computation

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agofeat(reforcexy): integrate warm start and continuous HPO
Jérôme Benoit [Tue, 30 Sep 2025 19:41:17 +0000 (21:41 +0200)] 
feat(reforcexy): integrate warm start and continuous HPO

And align tensorboard path with model training path namespace

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(qav3): cleanup namespace
Jérôme Benoit [Tue, 30 Sep 2025 14:37:57 +0000 (16:37 +0200)] 
refactor(qav3): cleanup namespace

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agofix(qav3): revert minimal_roi @property change
Jérôme Benoit [Tue, 30 Sep 2025 12:28:48 +0000 (14:28 +0200)] 
fix(qav3): revert minimal_roi @property change

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agofix(qav3): avoid config access to uninitialized properties
Jérôme Benoit [Tue, 30 Sep 2025 10:55:09 +0000 (12:55 +0200)] 
fix(qav3): avoid config access to uninitialized properties

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agoperf(qav3): tune fit_live_predictions_candles
Jérôme Benoit [Tue, 30 Sep 2025 10:46:13 +0000 (12:46 +0200)] 
perf(qav3): tune fit_live_predictions_candles

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(qav3): factor out callable unique hash building
Jérôme Benoit [Tue, 30 Sep 2025 09:40:06 +0000 (11:40 +0200)] 
refactor(qav3): factor out callable unique hash building

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agoperf(qav3): tune reversal confirmation
Jérôme Benoit [Tue, 30 Sep 2025 08:41:33 +0000 (10:41 +0200)] 
perf(qav3): tune reversal confirmation

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agofix(reforcexy): ensure env parent class close() is called
Jérôme Benoit [Mon, 29 Sep 2025 19:19:40 +0000 (21:19 +0200)] 
fix(reforcexy): ensure env parent class close() is called

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(reforcexy): cleanup spot support
Jérôme Benoit [Mon, 29 Sep 2025 17:46:05 +0000 (19:46 +0200)] 
refactor(reforcexy): cleanup spot support

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agofix(reforcexy): ensure model expiracy triggers open trades close
Jérôme Benoit [Mon, 29 Sep 2025 16:56:11 +0000 (18:56 +0200)] 
fix(reforcexy): ensure model expiracy triggers open trades close

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(qav3): kmeans/kmedoids code cleanups
Jérôme Benoit [Mon, 29 Sep 2025 16:08:01 +0000 (18:08 +0200)] 
refactor(qav3): kmeans/kmedoids code cleanups

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(qav3): cleanup kmedoids integration
Jérôme Benoit [Mon, 29 Sep 2025 15:39:02 +0000 (17:39 +0200)] 
refactor(qav3): cleanup kmedoids integration

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agochore(reforcexy): configuration template alignment
Jérôme Benoit [Sun, 28 Sep 2025 23:27:36 +0000 (01:27 +0200)] 
chore(reforcexy): configuration template alignment

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agoperf(qav3): slighly improve time complexity at pivots labeling
Jérôme Benoit [Sun, 28 Sep 2025 22:01:04 +0000 (00:01 +0200)] 
perf(qav3): slighly improve time complexity at pivots labeling

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agoperf(qav3): improve _get_n_clusters heuristic
Jérôme Benoit [Sun, 28 Sep 2025 21:06:41 +0000 (23:06 +0200)] 
perf(qav3): improve _get_n_clusters heuristic

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agochore(qav3): bump version
Jérôme Benoit [Sun, 28 Sep 2025 20:42:39 +0000 (22:42 +0200)] 
chore(qav3): bump version

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agofix(qav3): callback throttling per pair and callback once per candle
Jérôme Benoit [Sun, 28 Sep 2025 20:41:33 +0000 (22:41 +0200)] 
fix(qav3): callback throttling per pair and callback once per candle

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(qav3): cleanup MO _get_n_clusters
Jérôme Benoit [Sun, 28 Sep 2025 18:49:01 +0000 (20:49 +0200)] 
refactor(qav3): cleanup MO _get_n_clusters

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agoperf(qav3): improve MO optimization support
Jérôme Benoit [Sun, 28 Sep 2025 18:31:38 +0000 (20:31 +0200)] 
perf(qav3): improve MO optimization support

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agochore(reforcexy): refine configuration template
Jérôme Benoit [Sun, 28 Sep 2025 14:33:47 +0000 (16:33 +0200)] 
chore(reforcexy): refine configuration template

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agoperf(reforcexy): optimize prediction code path
Jérôme Benoit [Sun, 28 Sep 2025 14:00:21 +0000 (16:00 +0200)] 
perf(reforcexy): optimize prediction code path

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(reforcexy): refine log messages
Jérôme Benoit [Sat, 27 Sep 2025 19:34:26 +0000 (21:34 +0200)] 
refactor(reforcexy): refine log messages

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(reforcexy): improve logging
Jérôme Benoit [Sat, 27 Sep 2025 19:15:32 +0000 (21:15 +0200)] 
refactor(reforcexy): improve logging

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agoperf(reforcexy): make evaluation lighter
Jérôme Benoit [Sat, 27 Sep 2025 18:56:45 +0000 (20:56 +0200)] 
perf(reforcexy): make evaluation lighter

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agofix: typo in docker-upgrade.sh
Jérôme Benoit [Sat, 27 Sep 2025 15:33:46 +0000 (17:33 +0200)] 
fix: typo in docker-upgrade.sh

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agofix(reforcexy): ensure padding frame is init with zeros
Jérôme Benoit [Sat, 27 Sep 2025 15:29:28 +0000 (17:29 +0200)] 
fix(reforcexy): ensure padding frame is init with zeros

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(reforcexy): remove unneeded intermediate variable
Jérôme Benoit [Fri, 26 Sep 2025 23:26:56 +0000 (01:26 +0200)] 
refactor(reforcexy): remove unneeded intermediate variable

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(reforcexy): remove blank line
Jérôme Benoit [Fri, 26 Sep 2025 23:23:48 +0000 (01:23 +0200)] 
refactor(reforcexy): remove blank line

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(reforcexy): add more reward tunables
Jérôme Benoit [Fri, 26 Sep 2025 23:15:39 +0000 (01:15 +0200)] 
refactor(reforcexy): add more reward tunables

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(reforcexy): remove unneeded type casting
Jérôme Benoit [Fri, 26 Sep 2025 19:50:06 +0000 (21:50 +0200)] 
refactor(reforcexy): remove unneeded type casting

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agoperf(reforcexy): ensure total timesteps is aligned to model rollout with
Jérôme Benoit [Fri, 26 Sep 2025 19:31:38 +0000 (21:31 +0200)] 
perf(reforcexy): ensure total timesteps is aligned to model rollout with
PPO

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agofix(reforcexy): spot support at model training
Jérôme Benoit [Fri, 26 Sep 2025 18:29:59 +0000 (20:29 +0200)] 
fix(reforcexy): spot support at model training

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(reforcexy): keep nested directory structure at hypertopt
Jérôme Benoit [Fri, 26 Sep 2025 11:55:38 +0000 (13:55 +0200)] 
refactor(reforcexy): keep nested directory structure at hypertopt

Failure at finding a model is unrelated

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agofix(reforce): try to make hyperopt outputs not resusable by FT
Jérôme Benoit [Fri, 26 Sep 2025 11:33:31 +0000 (13:33 +0200)] 
fix(reforce): try to make hyperopt outputs not resusable by FT

And hopefully still cleaned up

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(reforcexy): refine typing
Jérôme Benoit [Thu, 25 Sep 2025 19:49:48 +0000 (21:49 +0200)] 
refactor(reforcexy): refine typing

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(reforcexy): refine typing
Jérôme Benoit [Thu, 25 Sep 2025 19:40:04 +0000 (21:40 +0200)] 
refactor(reforcexy): refine typing

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agoperf(qav3): refine signal filter
Jérôme Benoit [Thu, 25 Sep 2025 15:31:30 +0000 (17:31 +0200)] 
perf(qav3): refine signal filter

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(qav3): ensure safe hyperband pruner configuration
Jérôme Benoit [Thu, 25 Sep 2025 09:32:39 +0000 (11:32 +0200)] 
refactor(qav3): ensure safe hyperband pruner configuration

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agofix(reforcexy): train frequency handling
Jérôme Benoit [Wed, 24 Sep 2025 22:09:18 +0000 (00:09 +0200)] 
fix(reforcexy): train frequency handling

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(reforcexy): refine typing
Jérôme Benoit [Wed, 24 Sep 2025 21:06:56 +0000 (23:06 +0200)] 
refactor(reforcexy): refine typing

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(reforcexy): cleanup train frequency handling
Jérôme Benoit [Wed, 24 Sep 2025 21:05:18 +0000 (23:05 +0200)] 
refactor(reforcexy): cleanup train frequency handling

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agofix(reforcexy): ensure penalty at holding trade with pnl target reached
Jérôme Benoit [Wed, 24 Sep 2025 20:06:32 +0000 (22:06 +0200)] 
fix(reforcexy): ensure penalty at holding trade with pnl target reached

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agorefactor(reforcexy): ensure trade efficiency logic is applied at holding
Jérôme Benoit [Wed, 24 Sep 2025 17:40:10 +0000 (19:40 +0200)] 
refactor(reforcexy): ensure trade efficiency logic is applied at holding

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agofix(reforcexy): gradient steps computation
Jérôme Benoit [Wed, 24 Sep 2025 16:11:48 +0000 (18:11 +0200)] 
fix(reforcexy): gradient steps computation

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agochore(qav3): bump versions
Jérôme Benoit [Wed, 24 Sep 2025 15:16:54 +0000 (17:16 +0200)] 
chore(qav3): bump versions

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
4 weeks agoperf(reforcexy): revamp rewarding logic, take 2
Jérôme Benoit [Wed, 24 Sep 2025 15:12:23 +0000 (17:12 +0200)] 
perf(reforcexy): revamp rewarding logic, take 2

the initial revamp was catastrophic

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(qav3): code formatting
Jérôme Benoit [Tue, 23 Sep 2025 21:45:58 +0000 (23:45 +0200)] 
refactor(qav3): code formatting

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofix(qav3): fix sklearn extra installation
Jérôme Benoit [Tue, 23 Sep 2025 21:44:18 +0000 (23:44 +0200)] 
fix(qav3): fix sklearn extra installation

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofeat(qav3): add kmedoids metric to MO trial selection
Jérôme Benoit [Tue, 23 Sep 2025 21:28:41 +0000 (23:28 +0200)] 
feat(qav3): add kmedoids metric to MO trial selection

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(qav3): factor out pairwise distance sums computation
Jérôme Benoit [Tue, 23 Sep 2025 20:50:43 +0000 (22:50 +0200)] 
refactor(qav3): factor out pairwise distance sums computation

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): remove unneeded helper
Jérôme Benoit [Tue, 23 Sep 2025 18:40:01 +0000 (20:40 +0200)] 
refactor(reforcexy): remove unneeded helper

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agoperf(reforcexy): inital revamp of rewarding logic
Jérôme Benoit [Tue, 23 Sep 2025 13:52:16 +0000 (15:52 +0200)] 
perf(reforcexy): inital revamp of rewarding logic

* more continous reward space for gradient friendlyness
* HH and LL based reward when position if open

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): rename build_train_freq -> _build_train_freq
Jérôme Benoit [Mon, 22 Sep 2025 11:44:46 +0000 (13:44 +0200)] 
refactor(reforcexy): rename build_train_freq -> _build_train_freq

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): factor out DQN train_freq handling
Jérôme Benoit [Mon, 22 Sep 2025 11:35:12 +0000 (13:35 +0200)] 
refactor(reforcexy): factor out DQN train_freq handling

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): refine variables and methods namespace
Jérôme Benoit [Sun, 21 Sep 2025 22:48:47 +0000 (00:48 +0200)] 
refactor(reforcexy): refine variables and methods namespace

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): cleanup training schedule handling code
Jérôme Benoit [Sun, 21 Sep 2025 22:31:36 +0000 (00:31 +0200)] 
refactor(reforcexy): cleanup training schedule handling code

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(qav3): refine weighted_close() signature
Jérôme Benoit [Sun, 21 Sep 2025 20:38:34 +0000 (22:38 +0200)] 
refactor(qav3): refine weighted_close() signature

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agochore: refine configuration
Jérôme Benoit [Sun, 21 Sep 2025 20:24:51 +0000 (22:24 +0200)] 
chore: refine configuration

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): ensure envs are created with consistent data
Jérôme Benoit [Sun, 21 Sep 2025 20:11:32 +0000 (22:11 +0200)] 
refactor(reforcexy): ensure envs are created with consistent data
     snapshots

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofix(reforcexy): evaluation frequency computation
Jérôme Benoit [Sun, 21 Sep 2025 17:03:48 +0000 (19:03 +0200)] 
fix(reforcexy): evaluation frequency computation

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): code formatting
Jérôme Benoit [Sun, 21 Sep 2025 14:41:06 +0000 (16:41 +0200)] 
refactor(reforcexy): code formatting

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): refine log message
Jérôme Benoit [Sun, 21 Sep 2025 14:40:31 +0000 (16:40 +0200)] 
refactor(reforcexy): refine log message

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): factor out safe logger helper
Jérôme Benoit [Sun, 21 Sep 2025 14:37:54 +0000 (16:37 +0200)] 
refactor(reforcexy): factor out safe logger helper

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofix: ensure different hyperopt seed per trial
Jérôme Benoit [Sun, 21 Sep 2025 12:31:54 +0000 (14:31 +0200)] 
fix: ensure different hyperopt seed per trial

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofix: ensure std on samples apply Bessel correction
Jérôme Benoit [Sun, 21 Sep 2025 10:09:35 +0000 (12:09 +0200)] 
fix: ensure std on samples apply Bessel correction

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): ensure HPO trial start is always logged
Jérôme Benoit [Sat, 20 Sep 2025 20:07:27 +0000 (22:07 +0200)] 
refactor(reforcexy): ensure HPO trial start is always logged

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofix(reforcexy): fix arguments order at get_callbacks()
Jérôme Benoit [Sat, 20 Sep 2025 19:39:13 +0000 (21:39 +0200)] 
fix(reforcexy): fix arguments order at get_callbacks()

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agoperf(reforcexy): isolate and seed HPO envs
Jérôme Benoit [Sat, 20 Sep 2025 19:24:08 +0000 (21:24 +0200)] 
perf(reforcexy): isolate and seed HPO envs

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agoperf(reforcexy): make MyRLEnv a non nested class
Jérôme Benoit [Sat, 20 Sep 2025 16:35:34 +0000 (18:35 +0200)] 
perf(reforcexy): make MyRLEnv a non nested class

also fix another serious logic issue at optuna integration coming from
the orginal implementation.

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofix(reforcexy): ensure thread safety everywhere
Jérôme Benoit [Fri, 19 Sep 2025 20:07:45 +0000 (22:07 +0200)] 
fix(reforcexy): ensure thread safety everywhere

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofix(reforcexy): ensure thread safety with multiprocessing
Jérôme Benoit [Fri, 19 Sep 2025 19:59:34 +0000 (21:59 +0200)] 
fix(reforcexy): ensure thread safety with multiprocessing

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): refine configuration template
Jérôme Benoit [Fri, 19 Sep 2025 19:53:58 +0000 (21:53 +0200)] 
refactor(reforcexy): refine configuration template

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): ensure position arg is normalized
Jérôme Benoit [Fri, 19 Sep 2025 19:18:44 +0000 (21:18 +0200)] 
refactor(reforcexy): ensure position arg is normalized

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofix(reforcexy): fix access to _total_timesteps property
Jérôme Benoit [Fri, 19 Sep 2025 18:55:31 +0000 (20:55 +0200)] 
fix(reforcexy): fix access to _total_timesteps property

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): add learning progress tracking to tensorboard
Jérôme Benoit [Fri, 19 Sep 2025 18:41:02 +0000 (20:41 +0200)] 
refactor(reforcexy): add learning progress tracking to tensorboard

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): factor out action validation and masking logic
Jérôme Benoit [Fri, 19 Sep 2025 17:44:05 +0000 (19:44 +0200)] 
refactor(reforcexy): factor out action validation and masking logic

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agoperf(qav3): less strict reversal confirmation
Jérôme Benoit [Fri, 19 Sep 2025 16:36:03 +0000 (18:36 +0200)] 
perf(qav3): less strict reversal confirmation

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): remove dead code
Jérôme Benoit [Fri, 19 Sep 2025 16:03:49 +0000 (18:03 +0200)] 
refactor(reforcexy): remove dead code

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agoperf(qav3): fine tune labeling optuna optimization
Jérôme Benoit [Fri, 19 Sep 2025 14:55:53 +0000 (16:55 +0200)] 
perf(qav3): fine tune labeling optuna optimization

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): refine log message
Jérôme Benoit [Fri, 19 Sep 2025 14:37:11 +0000 (16:37 +0200)] 
refactor(reforcexy): refine log message

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): sensible variable name
Jérôme Benoit [Fri, 19 Sep 2025 13:54:27 +0000 (15:54 +0200)] 
refactor(reforcexy): sensible variable name

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofix(reforcexy): proper action masking implementation
Jérôme Benoit [Fri, 19 Sep 2025 13:48:00 +0000 (15:48 +0200)] 
fix(reforcexy): proper action masking implementation

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): rename variable
Jérôme Benoit [Fri, 19 Sep 2025 00:04:38 +0000 (02:04 +0200)] 
refactor(reforcexy): rename variable

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofix(reforcexy): fix get_most_recent_max_pnl() semantic
Jérôme Benoit [Thu, 18 Sep 2025 23:48:29 +0000 (01:48 +0200)] 
fix(reforcexy): fix get_most_recent_max_pnl() semantic

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agoperf(reforcexy): refine optuna trial pruning condition
Jérôme Benoit [Thu, 18 Sep 2025 21:09:23 +0000 (23:09 +0200)] 
perf(reforcexy): refine optuna trial pruning condition

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofeat(reforcexy): subprocvec support
Jérôme Benoit [Thu, 18 Sep 2025 20:09:48 +0000 (22:09 +0200)] 
feat(reforcexy): subprocvec support

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofix(reforcexy): properly separate hyperopt data from training data
Jérôme Benoit [Thu, 18 Sep 2025 13:27:23 +0000 (15:27 +0200)] 
fix(reforcexy): properly separate hyperopt data from training data

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofix(reforcexy): double evaluation during hyperopt
Jérôme Benoit [Thu, 18 Sep 2025 10:39:52 +0000 (12:39 +0200)] 
fix(reforcexy): double evaluation during hyperopt

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(qav3): properly integrate knn in MO trial selection
Jérôme Benoit [Thu, 18 Sep 2025 10:02:46 +0000 (12:02 +0200)] 
refactor(qav3): properly integrate knn in MO trial selection

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): cleanup frame stacking implementation
Jérôme Benoit [Wed, 17 Sep 2025 18:38:11 +0000 (20:38 +0200)] 
refactor(reforcexy): cleanup frame stacking implementation

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): align variable name
Jérôme Benoit [Wed, 17 Sep 2025 14:34:49 +0000 (16:34 +0200)] 
refactor(reforcexy): align variable name

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agofix(qav3): default value for pivot labeling
Jérôme Benoit [Wed, 17 Sep 2025 13:55:23 +0000 (15:55 +0200)] 
fix(qav3): default value for pivot labeling

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>
5 weeks agorefactor(reforcexy): filter properly float conversion errors
Jérôme Benoit [Wed, 17 Sep 2025 13:48:00 +0000 (15:48 +0200)] 
refactor(reforcexy): filter properly float conversion errors

Signed-off-by: Jérôme Benoit <jerome.benoit@piment-noir.org>