Skip to content

Fix varlen ref_gdr chunk offsets after prepare_chunk_offsets return change #16

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
ReinforcedKnowledge wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Fix varlen ref_gdr chunk offsets after prepare_chunk_offsets return change #16

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 9 additions & 5 deletions tests/ref_gdr.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -203,7 +203,8 @@ def torch_chunk_gdr_fwd(

if cu_seqlens is not None:
vn = pack(vn, cu_seqlens)
h = pack(h, prepare_chunk_offsets(cu_seqlens, chunk_size))
chunk_offsets, _ = prepare_chunk_offsets(cu_seqlens, chunk_size)
h = pack(h, chunk_offsets)

return h, vn, last_state

Expand All @@ -223,7 +224,8 @@ def torch_chunk_o_fwd(
k = unpack(k, cu_seqlens)
v = unpack(v, cu_seqlens)
g = unpack(g, cu_seqlens)
h = unpack(h, prepare_chunk_offsets(cu_seqlens, chunk_size))
chunk_offsets, _ = prepare_chunk_offsets(cu_seqlens, chunk_size)
h = unpack(h, chunk_offsets)

batch_size, num_tokens, num_k_heads, head_dim_k = k.shape
_, _, num_v_heads, head_dim_v = v.shape
Expand Down Expand Up @@ -379,7 +381,8 @@ def torch_chunk_gdr_bwd(
dv = dv.reshape((batch_size, -1, num_v_heads, head_dim_v))[:, :num_tokens]
if cu_seqlens is not None:
dv = pack(dv, cu_seqlens)
dh = pack(dh, prepare_chunk_offsets(cu_seqlens, chunk_size))
chunk_offsets, _ = prepare_chunk_offsets(cu_seqlens, chunk_size)
dh = pack(dh, chunk_offsets)
return dh, dh0, dv


Expand All @@ -405,8 +408,9 @@ def torch_chunk_dqkwg_bwd(
g = unpack(g, cu_seqlens)
do = unpack(do, cu_seqlens)
dv = unpack(dv, cu_seqlens)
h = unpack(h, prepare_chunk_offsets(cu_seqlens, chunk_size))
dh = unpack(dh, prepare_chunk_offsets(cu_seqlens, chunk_size))
chunk_offsets, _ = prepare_chunk_offsets(cu_seqlens, chunk_size)
h = unpack(h, chunk_offsets)
dh = unpack(dh, chunk_offsets)

batch_size, num_tokens, num_k_heads, head_dim_k = k.shape
_, _, num_v_heads, head_dim_v = do.shape
Expand Down