apollo-research/rl-dynamics-gradient-sketches-toy-adam5e5

Gradient sketches of a toy RL run (Adam, lr 5e-5, rep 3) Random-projection sketches of per-rollout policy gradients from a small GRPO run in Apollo Research's RL-dynamics project (base model Qwen/Qwen3.5-4B with LoRA adapters, 129,859,584 trainable parameters, 20 training steps, 1,024 rollouts per step). Every gradient was projected with the same subsampled randomised Hadamard transform (SRHT) to k = 262,144 coordinates, so inner products between sketches approximate inner… See the full description on the dataset page: https://huggingface.co/datasets/apollo-research/rl-dynamics-gradient-sketch

<iframe src="https://silytics.si/embed/item/3a2861ab-b87f-4c7a-8c1b-f4eca6ae34a2" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 19 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.