PhD Student at Mila, Montreal
Ankesh Anand
Karma: 80
Any plans on evaluating RETRO (the retrieval augmented transformer from DeepMind) on TruthfulQA? I’m guessing it should perform similarly to WebGPT but would be nice to get a concrete number.
Great post! I think you might wanna emphasize just how crucial ReAnalyse is for data-efficiency (the default MuZero is quite sample in-efficient), and how the reanalyse-ratio can be tuned easily for any data budget using a log-linear scaling law. You can also interpret the off-policy correction thing as running ReAnalyse twice, so my TL;DR of EfficientZero would be “MuZero ReAnalyse + SPR”.
Regarding contrastive vs SPR, I don’t think you would find a performance boost using a contrastive loss compared to SPR on Atari at least. We did an ablation for this in the SPR paper (Table 6, appendix). I suspect the reason contrastive works (slightly) better on Procgen is because of the procedural diversity there making negative examples much more informative.
Definitely agree about moving to multi-task test beds as the next frontier in RL. I also suspect we would see more non tabula-rasa RL methods, ones that start off with general-purpose pre-trained models or representations and then only do a tiny amount of fine-tuning on the actual RL task.
Thanks, glad you liked it, I really like the recent RL directions from OpenAI too! It would be interesting to see the use of model-based RL for the “RL as fine-tuning paradigm”: making large pre-trained models more aligned/goal-directed efficiently by simply searching over a reward function learned from humans.
I was eyeballing Figure 2 in the PPG paper and comparing it to our results on the full distribution (Table A.3).
PPO: ~0.25
PPG: ~0.52
MuZero: 0.68
MuZero+Reconstruction: 0.93
The Q-Learning baseline is a model-free control of MuZero. So it shares implementation details of MuZero (network architecture, replay ratio, training details etc.) while removing the model-based components of MuZero (details in sec A.2) . Some key differences you’d find vs a typical Q-learning implementation:
Larger network architectures: 10 block ResNet compared to a few conv layers in typical implementations.
Higher sample reuse: When using a reanalyse ratio of 0.95, both MuZero and Q-Learning use each replay buffer sample an average of 20 times. The target network is updated every 100 training steps.
Batch size of 1024 and some smaller details like using categorical reward and value predictions similar to MuZero.
We also have a small model-based component which predicts reward at next time step which lets us decompose the Q(s,a) into reward and value predictions just like MuZero.
I would guess larger networks + higher sample reuse have the biggest effect size compared to standard Q-learning implementations.
The ProcGen competition also might have used the easy difficulty mode compared to the hard difficulty mode used in our paper.
They do seem to cover SPR (an earlier version of SPR was called MPR). @flodorner If you do decide to update the plot, maybe you could update the label as well?
We do actually train/evaluate on the full distribution (See Figure 5 rightmost). MuZero+SSL versions (especially reconstruction) continue to be a lot more sample-efficient even in the full-distribution, and MuZero itself seems to be quite a bit more sample efficient than PPO/PPG.
Worth noting that they already use BERT in Search. https://blog.google/products/search/search-language-understanding-bert/
The raw neural network does use search during training though, and does not rely on search only during evaluation.
My main takeaway from Gato: If we can build specialized AI agents for 100s/1000s of tasks, it’s now pretty straightforward to make a general agent that can do it all in a single model. Just tokenize data from all the tasks and feed into a transformer.