distance weighted score implementation #106
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
total_goal_distance metric. For each agent: total_goal_distance = 2d_distance(initial_pos, goal_pos) distance_weighted_score = total_goal_distance * score The metrics shown in the dashboard are averages across agents.
| env->log.total_goal_distance += goal_distance; | ||
| if(e->reached_goal_this_episode && !e->collided_before_goal){ | ||
| env->log.score += 1.0f; | ||
| float goal_distance = ego_goal_distance_t0(e); |
There was a problem hiding this comment.
Interesting, do you know where this function is defined? I can't find it in the repo
Collaborator
Author
There was a problem hiding this comment.
line 311 of drive.h
| if(e->reached_goal_this_episode && !e->collided_before_goal){ | ||
| env->log.score += 1.0f; | ||
| float goal_distance = ego_goal_distance_t0(e); | ||
| env->log.distance_weighted_score += 1.0f * goal_distance; |
There was a problem hiding this comment.
I would be curious to see the distribution of distances to the goal, do you mind adding that? (can be done with wandb.Histogram)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Implemented the distance_weighted_score and total_goal_distance metric.
For each agent:
total_goal_distance = 2d_distance(initial_pos, goal_pos)
distance_weighted_score = total_goal_distance * score
The metrics collected are then averaged across agents as per current metric path. Example run: https://wandb.ai/emerge_/dc_pufferdrive/runs/5yyvzp3m?nw=nwuserrsavorgnanrs
The below image shows the "normalised distance of solved goals" (blue line), computed as:
distance_weighted_score / (score * total_goal_distance)
We're still investigating this metric, but an initial observation is that at convergence the solved goals have the same distance on average as the total population of goals, indicating that the unsolved goals might be "unsolved at random" with respect to the goal distance.