Skip to content

Add MoE and MLA remat policies #3414

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

Merged
copybara-service merged 3 commits into from
Mar 25, 2026
Merged

Add MoE and MLA remat policies #3414

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
d7fd385
Add MoE and MLA remat policies
abhinavgoel95 Mar 5, 2026
5109d24
Fix pyink formatting in attention_test.py
abhinavgoel95 Mar 24, 2026
0ac53b5
Remove moe_mlpwi for now
abhinavgoel95 Mar 24, 2026
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
6 changes: 6 additions & 0 deletions src/maxtext/configs/base.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -317,11 +317,17 @@ mlpwi: 'remat'
mlpwi_0: 'remat'
mlpwi_1: 'remat'
mlpwo: 'remat'
moe_mlpwi: 'remat'
moe_mlpwi_0: 'remat'
moe_mlpwi_1: 'remat'
moe_mlpwo: 'remat'
query_proj: 'remat'
key_proj: 'remat'
value_proj: 'remat'
qkv_proj: 'remat'
out_proj: 'remat'
query_wa_proj: 'remat'
kv_wa_proj: 'remat'
mla_q: 'remat'
mla_kv: 'remat'
attention_out: 'remat'
Expand Down
6 changes: 6 additions & 0 deletions src/maxtext/configs/pyconfig_deprecated.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -516,9 +516,15 @@ def validate_and_assign_remat_tensors(keys):
"mlpwi_0",
"mlpwi_1",
"mlpwo",
"moe_mlpwi",
"moe_mlpwi_0",
"moe_mlpwi_1",
"moe_mlpwo",
"query_proj",
"key_proj",
"value_proj",
"query_wa_proj",
"kv_wa_proj",
"out_proj",
]
assert keys["decoder_layer_input"] != "remat", "Cannot remeterialize this tensor with scan_layers=True"
Expand Down
24 changes: 24 additions & 0 deletions src/maxtext/configs/types.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -896,9 +896,33 @@ class RematAndOffload(BaseModel):
RematLocation.REMAT,
description="Remat policy for the second MLP layer's output.",
)
moe_mlpwi: RematLocation = Field(
RematLocation.REMAT,
description="Remat policy for the first MoE layer's intermediate output.",
)
moe_mlpwi_0: RematLocation = Field(
RematLocation.REMAT,
description="Remat policy for the first part of a gated MoE's output.",
)
moe_mlpwi_1: RematLocation = Field(
RematLocation.REMAT,
description="Remat policy for the second part of a gated MoE's output.",
)
moe_mlpwo: RematLocation = Field(
RematLocation.REMAT,
description="Remat policy for the second MoE layer's output.",
)
query_proj: RematLocation = Field(RematLocation.REMAT, description="Remat policy for the query projection.")
key_proj: RematLocation = Field(RematLocation.REMAT, description="Remat policy for the key projection.")
value_proj: RematLocation = Field(RematLocation.REMAT, description="Remat policy for the value projection.")
query_wa_proj: RematLocation = Field(
RematLocation.REMAT,
description="Remat policy for the MLA query weighted attention projection.",
)
kv_wa_proj: RematLocation = Field(
RematLocation.REMAT,
description="Remat policy for the MLA key and value weighted attention projection.",
)
qkv_proj: RematLocation = Field(RematLocation.REMAT, description="Remat policy for fused QKV projection.")
out_proj: RematLocation = Field(
RematLocation.REMAT,
Expand Down
2 changes: 2 additions & 0 deletions src/maxtext/layers/attention_mla.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -794,6 +794,7 @@ def mla_query_projection(
else:
# LoRA path
low_rank_q = self.wq_a(inputs_q, out_sharding=wqa_out_sharding) # [B, L, q_lora_rank]
low_rank_q = checkpoint_name(low_rank_q, "query_wa_proj")
low_rank_q = self.q_norm(low_rank_q) # RMSNorm on low rank
low_rank_q = checkpoint_name(low_rank_q, "mla_q")
q = self.wq_b(low_rank_q, out_sharding=query_sharding) # [B, L, n_heads, qk_head_dim]
Expand Down Expand Up @@ -933,6 +934,7 @@ def mla_kv_projection(self, inputs: Array, inputs_positions: Array, decoder_segm
wka_logical_name = (KV_BATCH, LENGTH_NO_EXP, KV_LORA_UP_PROJ)
wkva_out_sharding = create_sharding(self.mesh, wka_logical_name)
low_rank = self.wkv_a(inputs, out_sharding=wkva_out_sharding)
low_rank = checkpoint_name(low_rank, "kv_wa_proj")
low_rank_main, low_rank_rope = jnp.split(low_rank, [self.kv_lora_rank], axis=-1)
low_rank_main = self.kv_norm(low_rank_main)
low_rank_main = checkpoint_name(low_rank_main, "mla_kv")
Expand Down
18 changes: 9 additions & 9 deletions src/maxtext/layers/moe.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1274,7 +1274,7 @@ def get_active_sharding_axes(pspec_dim_axes, tensor_dim_index):
layer_w0 = jax.lax.psum(layer_w0, "tensor_transpose")
if self.config.mlp_bias:
layer_w0 = layer_w0 + w0_bias
layer_w0 = adc.checkpoint_name(layer_w0, "mlpwi_0")
Copy link
Copy Markdown
Collaborator

@RissyRan RissyRan Mar 14, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Heads up: this might affect all legacy TPU recipes/performance for MoE models. We should make an announcement after it gets merged. Thanks!

layer_w0 = adc.checkpoint_name(layer_w0, "moe_mlpwi_0")

layer_w1 = gmm_fn(
x,
Expand All @@ -1288,7 +1288,7 @@ def get_active_sharding_axes(pspec_dim_axes, tensor_dim_index):
layer_w1 = jax.lax.psum(layer_w1, "tensor_transpose")
if self.config.mlp_bias:
layer_w1 = layer_w1 + w1_bias
layer_w1 = adc.checkpoint_name(layer_w1, "mlpwi_1")
layer_w1 = adc.checkpoint_name(layer_w1, "moe_mlpwi_1")
intermediate_layer = self.apply_ffn_activation(layer_w0, layer_w1)

intermediate_output = gmm_fn(
Expand All @@ -1305,7 +1305,7 @@ def get_active_sharding_axes(pspec_dim_axes, tensor_dim_index):
)
if self.config.mlp_bias:
intermediate_output = intermediate_output + wo_bias
intermediate_output = adc.checkpoint_name(intermediate_output, "mlpwo")
intermediate_output = adc.checkpoint_name(intermediate_output, "moe_mlpwo")

if self.config.use_ring_of_experts:
# Set the outputs of tokens which were not processed to 0.
Expand Down Expand Up @@ -1860,7 +1860,7 @@ def dense_matmul(
layer_w0,
mlp_axis,
)
layer_w0 = adc.checkpoint_name(layer_w0, "mlpwi_0")
layer_w0 = adc.checkpoint_name(layer_w0, "moe_mlpwi_0")
with jax.named_scope("wi_1"):
w1_kernel_axes = ("exp", None, "mlp")
w1_kernel = self.maybe_all_gather_kernel_weight_in_expert_parallelism(w1_kernel, w1_kernel_axes)
Expand All @@ -1876,7 +1876,7 @@ def dense_matmul(
layer_w1,
mlp_axis,
)
layer_w1 = adc.checkpoint_name(layer_w1, "mlpwi_1")
layer_w1 = adc.checkpoint_name(layer_w1, "moe_mlpwi_1")
layer_multiply = self.apply_ffn_activation(layer_w0, layer_w1)
with jax.named_scope("wo"):
wo_kernel_axes = ("exp", "mlp", None)
Expand All @@ -1902,7 +1902,7 @@ def dense_matmul(
"activation_embed",
),
)
intermediate_layer = adc.checkpoint_name(intermediate_layer, "mlpwo")
intermediate_layer = adc.checkpoint_name(intermediate_layer, "moe_mlpwo")
with jax.named_scope("combine"):
# Matmul & element wise operation
output = self.get_einsum(rhs_mesh_axes=mask_axes, einsum_name=COMBINE)(
Expand Down Expand Up @@ -1931,7 +1931,7 @@ def dense_matmul(
layer_w0 = layer_w0 + w0_bias[None, None, :, :]
if self.config.activations_in_float32:
layer_w0 = layer_w0.astype(jnp.float32)
layer_w0 = adc.checkpoint_name(layer_w0, "mlpwi_0")
layer_w0 = adc.checkpoint_name(layer_w0, "moe_mlpwi_0")
with jax.named_scope("wi_1"):
layer_w1 = self.get_einsum(rhs_mesh_axes=self.wi_kernel_axes)(
"BSM,EMH -> BSEH", inputs, w1_kernel, precision=matmul_precision
Expand All @@ -1940,7 +1940,7 @@ def dense_matmul(
layer_w1 = layer_w1 + w1_bias[None, None, :, :]
if self.config.activations_in_float32:
layer_w1 = layer_w1.astype(jnp.float32)
layer_w1 = adc.checkpoint_name(layer_w1, "mlpwi_1")
layer_w1 = adc.checkpoint_name(layer_w1, "moe_mlpwi_1")
layer_multiply = self.apply_ffn_activation(layer_w0, layer_w1)

with jax.named_scope("wo"):
Expand All @@ -1954,7 +1954,7 @@ def dense_matmul(
intermediate_layer = intermediate_layer + wo_bias[None, None, :, :]
if self.config.activations_in_float32:
intermediate_layer = intermediate_layer.astype(jnp.float32)
intermediate_layer = adc.checkpoint_name(intermediate_layer, "mlpwo")
intermediate_layer = adc.checkpoint_name(intermediate_layer, "moe_mlpwo")
with jax.named_scope("weight_sum"):
if is_llama4_decoder_layer:
weights = self.reshape_and_update_weights(jnp.ones_like(top_k_weights), top_k_indices)
Expand Down
Loading