Skip to main content

Rotary Test 3

150M param model on OWT2 with learned embeddings (blue) vs. rotary embeddings (green) vs. rpe (brown) vs. rpe with caching (peach)
Created on April 15|Last edited on April 15

Section 1


50k100k150kStep34
group: c8h5RaxCLDDWmEdewhGjTs
group: EaSTz5R9RzXz9QjfVHBWcc
group: dq2tfRJtyHb2L2hM2yn9zB
group: TWYUDCsHVajhpJBxoiAbzc
50k100k150kStep345678910
pos emb: rpe group: c8h5RaxCLDDWmEdewhGjTs
pos emb: rpe group: EaSTz5R9RzXz9QjfVHBWcc
pos emb: rotary group: dq2tfRJtyHb2L2hM2yn9zB
pos emb: learned group: TWYUDCsHVajhpJBxoiAbzc
Run set
4



Run set
4