AllenNeuralDynamics / Aind.Behavior.VrForaging

Task implemention of VR Foraging experiment
https://allenneuraldynamics.github.io/Aind.Behavior.VrForaging/
MIT License
2 stars 0 forks source link

Issues while testing 0.5.0 release #313

Closed tiffanyona closed 1 day ago

tiffanyona commented 2 days ago
  1. The VR display lags a bit, it's not as fluid as before.
  2. The name of the mouse and task at the bottom is cropped (independently of the maximization of the window
  3. The naming convention for the data folders changed, right?
  4. end experiment moves the spout vertically (Z) instead of backwards (y1 and y2)
  5. I am not sure what should I expect after I press end experiment. It basically stops the workflow but hangs everything else with no message in terminal information of the state? I only tried with the launcher, I will try it in VS, maybe it is a bit more informative.
tiffanyona commented 2 days ago
  1. Lookup table not behaving as intented. test_PR.json
def LookUpDepletion(
    amount_drop: int = 5,
    available_water: int = 50,
    max_p=0.9,
    stop_duration: float = 0.5,
    delay_mean: float = 0.5,
    option: str = "single",
):
    if option == "delayed":
        lut_values = [0,0,max_p, max_p, max_p] + list(np.repeat(0, 45))
    elif option == "single":
        lut_values = [max_p] + list(np.repeat(0, 49))
    elif option == "no_reward":
        lut_values = list(np.repeat(0, 50))
    return vr_task_logic.RewardSpecification(
        operant_logic=OperantLogicHelper(stop_duration=stop_duration, is_operant=False),
        delay=NormalDistributionHelper(delay_mean, 0.15, 0.0, 1),
        reward_function=vr_task_logic.PatchRewardFunction(
            amount=vr_task_logic.RewardFunction(vr_task_logic.ConstantFunction(value=amount_drop)),
            probability=vr_task_logic.RewardFunction(vr_task_logic.LookupTableFunction(
                lut_keys=list(np.arange(0,50)), lut_values=lut_values
                ),
            ),
            available=vr_task_logic.RewardFunction(vr_task_logic.ConstantFunction(value=available_water)),
            depletion_rule=vr_task_logic.DepletionRule.ON_CHOICE,
        ),
    )
bruno-f-cruz commented 2 days ago
  1. After trying it at the rig we could not reproduced. I will close this issue for now. If it keeps happening feel free to open a new issue for this specific problem.
  2. Fixed in 1f59004f4f73f3f5732692a69352fe88c94a583f
  3. Already documented in the PR (#289 and in Services docs). If something doesn't match, it should be considered a bug and reported.
  4. Fixed in 9f3c31f06f98ff718d0aabef2cff06e20acb765b
  5. You will need to close the bonsai window. In order to do everything automatic, bonsai would need to run in no-editor mode which we agreed we did not want to do for now.
  6. Fixed in d2a61a0033a6b961f2c5239f54df612340e64468