add methods

[Andron00e]

• SOAP
• Muon (add schedules)
• Shampoo (only DistributedShampoo)
• Adam-mini
• Lion
• Sophia
• AdEMAMix
• Schedule-Free
• Adafactor
• Adalayer
• Signum, signSGD
• AdaHessian
• Prodigy
• SGDF
• ADOPT
• ...

[Andron00e]

add schedules also | link

upd: has been added via this commit

[Andron00e]

some problems with installation of the lates version of schedulefree, so I added this manually
see: https://github.com/epfml/llm-baselines/blob/soap/src/optim/schedulefree.py

[martinjaggi]

is there a pull request for this? would be nice to collaborate

[Andron00e]

is there a pull request for this? would be nice to collaborate

hi, we are deploying it to the soap branch together with @mpagli

[Andron00e]

a useful settings:

• anything but sgd
• ...

[Andron00e]

Adam-mini Note

I use model.named_parameters() for Adam-mini instead of group_specs, therefore in main.py it looks like:

  elif args.opt == "adam-mini":
      opt = Adam_mini(
          device=args.device,
          world_size=args.world_size,
          named_parameters=model.named_parameters(),  # check
          lr=args.lr,
          betas=(args.beta1, args.beta2),
          weight_decay=args.weight_decay,
          model_sharding=args.model_sharding,
          dim=args.n_embd,
          n_heads=args.n_head,
          n_kv_heads=args.n_kv_head,
          verbose=args.adam_mini_verbose,
      )

TODO: update partitions names

[kylematoba]

hi, I'll add sophia and adafactor.

[Andron00e]

hi, I'll add sophia and adafactor.

Hello! Super, just develop this in your branch and then PR to soap. I am a bit overloaded these days, but wanted to try Sophia also

Note: in official repository, they do not show SophiaH (with Hutchinson's preconditioner), only SophiaG. We want to have both methods here. SophiaH is nicely implemented in optax for now, but its not so hard to write in PyToch, see: this link

Thx)

[kylematoba]

hi, Bristen is back early, so I'll get back to that.

I did some research on Sophia, though, main findings:

• The official implementation of SophiaG makes some weird choices, described here https://web.stanford.edu/class/archive/cs/cs224n/cs224n.1244/final-projects/CaiaMaiCostelloJasonDanielLazar.pdf.
• The levanter implementation does not have SophiaG, only SophiaH.
• There's a quite readable Julia implementation of SophiaH here https://github.com/SciML/Optimization.jl/blob/master/src/sophia.jl.

Adafactor is simple, it's already close to being released officially, see pytorch/pytorch#129905.

When I get some time next I'll return to this if you haven't.

[martinjaggi]

muon optimizer should also be a good one to add. i think @doikov might be interested in that one too:
https://x.com/Yuchenj_UW/status/1846964136204173318

[martinjaggi]

once we have a handful, we'll have a nice benchmark collection for LLM optimizers, probably worth a small writeup soon

[Andron00e]

muon optimizer should also be a good one to add. i think @doikov might be interested in that one too: https://x.com/Yuchenj_UW/status/1846964136204173318

yes, i am working on that.
already have some test runs of the Muon. but, again, it is hard to deduce when batch size is less than 0.5M tokens

btw
an interesting exercise – to try this new muon/soap/whatever on the banana function :)

[Andron00e]

hi, Bristen is back early, so I'll get back to that.

I did some research on Sophia, though, main findings:

• The official implementation of SophiaG makes some weird choices, described here https://web.stanford.edu/class/archive/cs/cs224n/cs224n.1244/final-projects/CaiaMaiCostelloJasonDanielLazar.pdf.
• The levanter implementation does not have SophiaG, only SophiaH.
• There's a quite readable Julia implementation of SophiaH here https://github.com/SciML/Optimization.jl/blob/master/src/sophia.jl.

Adafactor is simple, it's already close to being released officially, see pytorch/pytorch#129905.

When I get some time next I'll return to this if you haven't.

I mean, for the official version of SophaiG, you may just look at the paper's repo: https://github.com/Liuhong99/Sophia