-
Why is the performance so poor? Is there a bug, or is the model itself just this poor in performance?
-
### 🚀 The feature, motivation and pitch
PPO and a number of other LLM fine-tuning techniques require autoregressive generation as part of the training process. When using vLLM to speed up the autor…
-
They claim lookahead decoding provides a 1.5~2x decoding speedup without a speculative model.
Blog post: https://lmsys.org/blog/2023-11-21-lookahead-decoding/
Twitter thread: https://twitter.com/l…
-
### 🚀 The feature, motivation and pitch
I calculated in non-autoregressive tts, asr, machine-translation using transformer with convolutional network. Machine learning was well carried out. So, I …
-
(I never looked at this)
What model can we use instead of MultivariateOLS when we have changing variance, cov_resid?
example: I'm using the growthrate (diff log) of the macro dataset in VAR and …
-
It would be great if `ContinuousTransformerWrapper` supported `return_mems` in the forward pass.
Thank you for the awesome repo!
Remarkably, it all works with `torch.onnx.export()`!
-
After training the network on a dataset of music, when using the inference script with my trained model and comparing the results with the original audio files, I noticed that after the 30 second mark…
-
This looks like a very interesting approach for general nonlinear/non-normal time series models, including point processes
(mentions Poisson ARMA type models as special case, but most applications ar…
-
When i use run_union.py do predict-tokens task
Input texts are:
1
An alien race encounters the most terrifying predator imaginable . A lone , unarmed human . The last time they 'd come ... they '…
-
Thank you for the latest version of the code release. When I actually trained and used different sampling strategies, I found that the effect of pyramid sampling is not as good as full_sequence_sampli…