-
Notifications
You must be signed in to change notification settings - Fork 1.9k
Weight vector
arielf edited this page Jun 29, 2012
·
9 revisions
VW's weight vector has (2^b) weights (where (b) is specified by the -b
option) and each example's features are hash to an index in ([0,2^b-1]). The weight vector is also used to store other vectors needed by more sophisticated learning algorithms, such as the conjugate gradient method (--conjugate_gradient
), or adaptive gradient descent (--adaptive
and/or --exact_adaptive_norm
).
When more than one vector is stored in the same global (2^b) space, every hash-value slot will store two (or more) "weights" so the slot hash value is first integer divided (hash_value / N) to store N values per slot. You may want to consider increasing the -b
option value to avoid hash-collisions in these cases.
- Home
- First Steps
- Input
- Command line arguments
- Model saving and loading
- Controlling VW's output
- Audit
- Algorithm details
- Awesome Vowpal Wabbit
- Learning algorithm
- Learning to Search subsystem
- Loss functions
- What is a learner?
- Docker image
- Model merging
- Evaluation of exploration algorithms
- Reductions
- Contextual Bandit algorithms
- Contextual Bandit Exploration with SquareCB
- Contextual Bandit Zeroth Order Optimization
- Conditional Contextual Bandit
- Slates
- CATS, CATS-pdf for Continuous Actions
- Automl
- Epsilon Decay
- Warm starting contextual bandits
- Efficient Second Order Online Learning
- Latent Dirichlet Allocation
- VW Reductions Workflows
- Interaction Grounded Learning
- CB with Large Action Spaces
- CB with Graph Feedback
- FreeGrad
- Marginal
- Active Learning
- Eigen Memory Trees (EMT)
- Element-wise interaction
- Bindings
-
Examples
- Logged Contextual Bandit example
- One Against All (oaa) multi class example
- Weighted All Pairs (wap) multi class example
- Cost Sensitive One Against All (csoaa) multi class example
- Multiclass classification
- Error Correcting Tournament (ect) multi class example
- Malicious URL example
- Daemon example
- Matrix factorization example
- Rcv1 example
- Truncated gradient descent example
- Scripts
- Implement your own joint prediction model
- Predicting probabilities
- murmur2 vs murmur3
- Weight vector
- Matching Label and Prediction Types Between Reductions
- Zhen's Presentation Slides on enhancements to vw
- EZExample Archive
- Design Documents
- Contribute: