-
微博内容精选
-
Song Y, Ermon S. [Generative modeling by estimating gradients of the data distribution](https://proceedings.neurips.cc/paper/2019/file/3001ef257407d5a371a96dcd947c7d93-Paper.pdf).
Yang Song. [Generat…
-
I have an implementation of Annealed Importance Sampling (AIS) at https://github.com/treigerm/AnnealedIS.jl and @ParadaCarleton kindly suggested that I could merge it into an official Turing package a…
-
I'm currently using a pure-numpy implementation of Annealed Importance Sampling (for binary multivariate RVs), and I'm trying to improve performance by porting the code to jax. I'm benchmarking the ja…
-
This is a follow-up to
https://github.com/uber/pyro/issues/568
As was recommended in the discussion of the above issue, even if we can model them with delta or normal with small variance samples,…
innuo updated
3 years ago
-
In:
tensorflow_probability/g3doc/api_docs/python/tfp/mcmc/sample_annealed_importance_chain.md
Line 93:
```
proposal = tfd.MultivatiateNormalDiag(
loc=tf.zeros([dims], dtype=dtype))
```
S…
-
Hi,
first of all thanks for sharing the code!
I've been playing around with it, and I noticed that the function `TRBM_logZperbin_Annealed_Importance_Sampling` is not defined anywhere, but is cal…
-
Hi,
I've been playing with marginal likelihood estimation techniques for GP classification, and I've found some strange behavior with the EP calculations. I wrote my own annealed importance sampling c…
-
Currently, the geometry proposal `x_new ~ p(new | old)` (the "geometry" step below) and computation of the Metropolis-Hastings ratio `p(old | new) / p(new | old)` (the "ratio" step below) is done in a…
-
Possible causes:
1. Mean field is a bad training method
2. Mean field is fine for training, but it understates the true likelihood during testing
3. I'm doing something wrong
Currently leaning toward…