[DeepSeek-V4] Implement Compressed Attention Layers by parambole · Pull Request #3866 · AI-Hypercomputer/maxtext

parambole · 2026-05-11T17:48:57Z

Description

Implement compressed attention mechanisms and indexer modules required for DeepSeek-V4 integration into MaxText:

CSACompressor & HCACompressor: Long-range attention compressors supporting causal block bias and YaRN frequency scaling decoupling.
LightningIndexer: Memory-efficient indexer module implementing sentinel masking and dynamic RoPE scaling.
Configuration: Register attention compression hyperparameters (compress_ratios, index_head_dim, sliding_window) in types.py and base.yml.
Unit test suite (tests/unit/deepseek_v4_vs_reference_test.py) validating attention compression parity against PyTorch reference implementations at atol=1e-5, rtol=1e-5.

Tests

Tested on CPU

pytest  tests/unit/deepseek_v4_vs_reference_test.py

======================= 10 passed, 10 warnings in 20.42s =======================
tests/unit/deepseek_v4_vs_reference_test.py ..........                   [100%]

Checklist

Before submitting this PR, please make sure (put X in square brackets):

I have performed a self-review of my code. For an optional AI review, add the gemini-review label.
I have necessary comments in my code, particularly in hard-to-understand areas.
I have run end-to-end tests tests and provided workload links above if applicable.
I have made or will make corresponding changes to the doc if needed, including adding new documentation pages to the relevant Table of Contents (toctree directive) as explained in our documentation.

codecov · 2026-05-11T17:56:22Z

Codecov Report

❌ Patch coverage is 0% with 262 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/maxtext/layers/attention_compressed.py	0.00%	262 Missing ⚠️

📢 Thoughts on this report? Let us know!

…ghtningIndexer) Implement compressed attention mechanisms and indexer modules for DeepSeek-V4 integration into MaxText: - CSACompressor & HCACompressor: Long-range attention compressors supporting causal block bias and YaRN frequency scaling decoupling. - LightningIndexer: Memory-efficient indexer module implementing sentinel masking and dynamic RoPE scaling. - Configuration: Register attention compression hyperparameters (compress_ratios, index_head_dim, sliding_window) in types.py and base.yml. - Parity verification: Extended unit test suite (deepseek_v4_vs_reference_test.py) validating attention compression parity against PyTorch reference implementations at atol=1e-5, rtol=1e-5.

github-actions · 2026-05-14T18:32:33Z

🤖 Hi @parambole, I've received your request, and I'm working on it now! You can track my progress in the logs for more details.

github-actions · 2026-05-14T20:37:57Z

🤖 Hi @parambole, I've received your request, and I'm working on it now! You can track my progress in the logs for more details.

github-actions

## 📋 Review Summary

This PR implements the core compressed attention layers for DeepSeek-V4 integration, including HCA, CSA, and the Lightning Indexer. The implementation is technically sound, follows established patterns in MaxText, and includes comprehensive parity tests against PyTorch.

🔍 General Feedback

Efficiency: The main coordinator block uses jnp.repeat for broadcasting MQA keys/values, which is memory-intensive. Switching to jnp.einsum broadcasting is recommended.
Typo: A minor typo swaped was found in the indexer module.
Config: Ensure compress_ratios is properly documented as a required list when using these attention variants to avoid runtime IndexError.

github-actions · 2026-05-14T20:40:24Z

+
+    # Compute attention logits
+    # logits: [B, H, S, S_kv]
+    logits = jnp.einsum("bhsd, bhkd -> bhsk", q, k, precision=self.config.matmul_precision) * self.scaling


🔴 Memory/Performance: `jnp.repeat` creates large intermediate tensors that are unnecessary given the shared MQA structure. You can use broadcasting in `jnp.einsum` to achieve the same result more efficiently.

Suggested change

logits = jnp.einsum("bhsd, bhkd -> bhsk", q, k, precision=self.config.matmul_precision) * self.scaling

# Broadcast key/value configurations to all heads using broadcasting in einsum

# k and v remain [B, 1, S_kv, D_head]

k = kv

v = kv

# Compute attention logits with head broadcasting: [B, H, S, S_kv]

logits = jnp.einsum("bhsd, b1kd -> bhsk", q, k, precision=self.config.matmul_precision) * self.scaling

github-actions · 2026-05-14T20:40:24Z

+
+    # Project attention weights onto values
+    # attn_output: [B, H, S, D_head]
+    attn_output = jnp.einsum("bhsk, bhkd -> bhsd", attn_weights, v, precision=self.config.matmul_precision)


🔴 Memory/Performance: Use broadcasting in `jnp.einsum` here as well to avoid the `jnp.repeat` from earlier.

Suggested change

attn_output = jnp.einsum("bhsk, bhkd -> bhsd", attn_weights, v, precision=self.config.matmul_precision)

# Project attention weights onto values with head broadcasting: [B, H, S, D_head]

attn_output = jnp.einsum("bhsk, b1kd -> bhsd", attn_weights, v, precision=self.config.matmul_precision)

github-actions · 2026-05-14T20:40:26Z

+    swaped_kv = jnp.swapaxes(compressed_kv, -1, -2)
+    swaped_kv = jnp.expand_dims(swaped_kv, axis=1)
+    # scores: [B, S, H, W]
+    scores = jnp.matmul(q, swaped_kv)


🟢 Typo: "swaped" should be "swapped".

Suggested change

scores = jnp.matmul(q, swaped_kv)

# swapped_kv: [B, 1, D_idx, W]

swapped_kv = jnp.swapaxes(compressed_kv, -1, -2)

swapped_kv = jnp.expand_dims(swapped_kv, axis=1)

# scores: [B, S, H, W]

scores = jnp.matmul(q, swapped_kv)

github-actions · 2026-05-14T20:40:27Z

+  """Configuration specific to DeepSeek-V4 stateless compressed attention layers."""
+
+  compress_rope_theta: float = Field(160000.0, description="Theta base frequency for long-range compressor layers.")
+  compress_ratios: list[int] = Field(


🟡 Configuration: The default `compress_ratios` is an empty list. Since `DeepSeekV4Attention` relies on this list having at least `layer_idx + 1` elements when using compressed layer types, this will cause an `IndexError` at runtime unless the user provides a full list. It might be better to provide a default or add a check with a clear error message.

RissyRan

I didn't see sharding annotations. It will be good we start to add some of them in this PR? i.e. starting with those weights.

RissyRan · 2026-05-16T04:01:16Z

+    return compressed_kv, block_bias
+
+
+class DeepSeekV4Indexer(nnx.Module):


Have you considered to re-use v3.2 Indexer?

Ref: doc

I agree with reusing. Can we at least have a BaseIndexer class to extract common logic (e.g., init, scoring)? Then v3.2 indexer and v4 indexer inherits from this.

Here is an example

class BaseIndexer(nnx.Module): """Base Lightning Indexer for Sparse Attention""" def __init__( self, config: Any, kernel_init: NdInitializer, quant: Optional[Quant] = None, weights_proj_in_features: Optional[int] = None, weights_proj_dtype: Optional[DType] = None, weights_proj_quant: Optional[Quant] = None, rngs: Optional[nnx.Rngs] = None, ): self.config = config self.dtype = config.dtype self.weight_dtype = config.weight_dtype self.n_heads = config.indexer_n_heads self.head_dim = config.indexer_head_dim self.indexer_topk = config.indexer_topk self.q_lora_rank = config.q_lora_rank self.softmax_scale = self.head_dim**-0.5 self.weights_scaling = self.n_heads**-0.5 # Query Projection: Latent Query -> Indexer Query self.wq_b = DenseGeneral( in_features_shape=self.q_lora_rank, out_features_shape=(self.n_heads, self.head_dim), axis=-1, kernel_init=kernel_init, kernel_axes=("q_lora", "q_heads", "kv"), dtype=self.dtype, weight_dtype=self.weight_dtype, quant=quant, matmul_precision=config.matmul_precision, shard_mode=config.shard_mode, rngs=rngs, ) wp_in_features = weights_proj_in_features if weights_proj_in_features is not None else config.emb_dim wp_dtype = weights_proj_dtype if weights_proj_dtype is not None else self.dtype # Projection: Input -> Importance Weights for Heads self.weights_proj = DenseGeneral( in_features_shape=wp_in_features, out_features_shape=self.n_heads, axis=-1, kernel_init=kernel_init, kernel_axes=("embed", "q_heads"), dtype=wp_dtype, weight_dtype=wp_dtype, quant=weights_proj_quant, matmul_precision=config.matmul_precision, shard_mode=config.shard_mode, rngs=rngs, ) def prepare_query(self, low_rank_q: jnp.ndarray, inputs_positions: jnp.ndarray, apply_partial_rope) -> jnp.ndarray: bsz, seqlen, _ = low_rank_q.shape q = self.wq_b(low_rank_q) q = q.reshape(bsz, seqlen, self.n_heads, self.head_dim) q = apply_partial_rope(q, inputs_positions=inputs_positions) return q def compute_topk(self, q: jnp.ndarray, k: jnp.ndarray, inputs_q: jnp.ndarray, mask: Optional[jnp.ndarray]) -> tuple[jnp.ndarray, jnp.ndarray]: logits = jnp.einsum("bthd, bsd -> btsh", q, k, precision=self.config.matmul_precision) logits = jax.nn.relu(logits) weights = self.weights_proj(inputs_q) weights = weights * self.weights_scaling * self.softmax_scale indexer_score = jnp.einsum("btsh, bth -> bts", logits, weights, precision=self.config.matmul_precision) if mask is not None: indexer_score += mask _, topk_indices = jax.lax.top_k(indexer_score, k=self.indexer_topk) return topk_indices, indexer_score

RissyRan · 2026-05-16T04:06:04Z

+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Compressed Attention layers and long-range compressors."""


For file name, may be compressed_attention.py as you mentioned in the comment here?

maybe it is influenced byattention_mla.py?

…ry expansion parameters for Trillium parity

shuningjin

Thanks for the implementation! Here are my high-level comments:

Please replace nnx.Linear with MaxText DenseGeneral
I recommend reusing some common logic from Indexer of V3.2, perhaps by abstracting into a base class
It unclear to me how "Additional Branch of Sliding Window Attention" from Section 2.3.3 in paper is implemented. Could you explain?

shuningjin · 2026-05-26T10:13:05Z

+index_head_dim: 128
+index_n_heads: 64
+index_topk: 512


These can be removed. Please reuse existing indexer config:

maxtext/src/maxtext/configs/base.yml

Lines 379 to 384 in 9d79b99

# DeepSeek Sparse Attention (DSA)

# deepseek3.2 introduces indexer in MLA

use_indexer: False

indexer_head_dim: 128

indexer_n_heads: 64

indexer_topk: 2048

shuningjin · 2026-05-26T10:24:07Z

+index_topk: 512
+o_groups: 8
+o_lora_rank: 1024
+sliding_window: 128


Can you make the name sliding_window more specific?

It can be easily confused with sliding_window_size for attention_type=local_sliding.

maxtext/src/maxtext/configs/base.yml

Lines 358 to 362 in 9d79b99

attention_type: 'global' # Supported attention_type: global, local_sliding, chunk, mla

share_kv_projections: False # Note: Not compatible with attention_type='mla'

attention_bias: False # If True, adds a learnable bias to the query, key, and value projections

attention_sink: False

sliding_window_size: 0

Where are we using config.sliding_window? I didn't see it in code.

shuningjin · 2026-05-26T10:32:20Z

    MlaAttention,
    MoBa,
    AttentionIndexer,
+    DeepSeekV4AttentionConfig,
    Llama4Attention,
    SplashAttention,
    PagedAttention,


nit: DeepSeekV4AttentionConfig -> DeepSeekV4Attention, to be consistent with other class in types.py.

shuningjin · 2026-05-26T10:34:40Z

+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Compressed Attention layers and long-range compressors."""


maybe it is influenced byattention_mla.py?

shuningjin · 2026-05-26T11:06:55Z

+    rope_theta = config.compress_rope_theta
+
+    # Linear projection of inputs to key/value representation
+    self.kv_proj = nnx.Linear(


For all occurrences of nnx.Linear in the PR, should be replaced with MaxText's DenseGeneral.

from maxtext.layers.linears import DenseGeneral

See these examples:

maxtext/src/maxtext/layers/attentions.py

Lines 655 to 668 in 4110d13

return DenseGeneral(

in_features_shape=self.convert_dense_general_inputs_shape(inputs_kv_shape),

out_features_shape=(self.num_kv_heads, self.head_dim),

axis=-1,

kernel_init=self.kernel_init,

kernel_axes=kernel_axes,

dtype=self.dtype,

weight_dtype=self.weight_dtype,

quant=self.quant,

shard_mode=self.config.shard_mode,

matmul_precision=self.config.matmul_precision,

use_bias=self.use_bias_in_projections,

rngs=self.rngs,

)

maxtext/src/maxtext/layers/attention_mla.py

Lines 716 to 743 in 4110d13

def _init_projections(self, inputs_q_shape: Tuple, inputs_kv_shape: Tuple) -> None:

"""Initializes the MLA-specific projections."""

# Assert required configuration parameters for MLA attention.

assert (

self.config.attention_type == AttentionType.MLA.value

), f"MLA requires MLA attention type {AttentionType.MLA.value}"

assert self.kv_lora_rank > 0, "KV LoRA rank must be > 0"

assert self.qk_nope_head_dim > 0, "QK NoPe head dim must be > 0"

assert self.qk_rope_head_dim > 0, "QK RoPE head dim must be > 0"

assert self.v_head_dim > 0, "V head dim must be > 0"

assert self.num_query_heads == self.num_kv_heads, "MLA requires equal number of query and kv heads"

assert not self.config.fused_qkv, "Fused QKV is not supported for MLA"

if self.q_lora_rank == 0:

# Standard Q projection (without LoRA).

self.query = DenseGeneral(

in_features_shape=self.config.emb_dim,

out_features_shape=(self.num_query_heads, self.qk_head_dim),

axis=-1,

kernel_init=self.kernel_init,

kernel_axes=("embed", "q_heads", "kv"),

dtype=self.dtype,

weight_dtype=self.weight_dtype,

quant=self.quant,

matmul_precision=self.config.matmul_precision,

shard_mode=self.config.shard_mode,

rngs=self.rngs,

)

shuningjin · 2026-05-26T14:09:27Z

+    self.attention_type = attention_type
+    self.num_heads = num_heads
+    self.head_dim = head_dim
+    self.sliding_window = config.sliding_window


where is this used? where is "Additional Branch of Sliding Window Attention" in Sec 2.3.3?

shuningjin · 2026-05-26T14:15:13Z

    self.layer_idx = layer_idx
    self.layer_type = config.layer_types[layer_idx]
+    # Sliding-only layers use the "main" (plain θ=10000) rope; CSA/HCA layers
+    # share the same yarn-scaled "compress" rope as their compressor.


Where is yarn incorporated? I didn't see it in #3865

shuningjin · 2026-05-26T14:21:49Z

+    # Projections for query extraction and low-rank compression
+    self.q_a_proj = nnx.Linear(


nit: The query low rank logic seems identical to v3. Maybe make the name consistent with v3:

q_a_proj -> wq_a q_b_proj -> wq_b

shuningjin · 2026-05-26T14:42:36Z

+      if attention_mask is not None:
+        # Pad 4D attention mask along trailing key/value sequence axis.
+        # # [B, 1, Q, S + W] -> [B, 1, Q, align(S + W, sa_block_kv)]
+        attention_mask = jnp.pad(attention_mask, ((0, 0), (0, 0), (0, 0), (0, pad_len)), constant_values=0.0)


pad with -jnp.inf instead of 0? or perhaps DEFAULT_MASK_VALUE

maxtext/src/maxtext/common/common_types.py

Lines 72 to 74 in 4110d13

# A large negative mask value is used for masking to ensure that the

# softmax function assigns an extremely low probability to the masked positions.

DEFAULT_MASK_VALUE = -0.7 * float(np.finfo(np.dtype("float32")).max)

shuningjin · 2026-05-26T14:43:29Z

+
+    # Reconcile structural block bias masks with runtime attention masks.
+    if attention_mask is not None:
+      if block_bias is not None:


What happens if attention_mask is None and block_bias is not None?

entrpn

LGTM. Please resolve other comments before merging.

parambole force-pushed the deepseek_v4_compressed_attention branch from 5f54827 to 07eb3e2 Compare May 11, 2026 19:39

parambole changed the base branch from deepseek_v4_core_primitives to dsv4-moe-routing-primitives May 11, 2026 20:29

parambole force-pushed the dsv4-moe-routing-primitives branch from 37ee811 to 31329c5 Compare May 11, 2026 20:38

parambole force-pushed the deepseek_v4_compressed_attention branch from 07eb3e2 to 4520166 Compare May 11, 2026 20:43

parambole force-pushed the dsv4-moe-routing-primitives branch from 31329c5 to 22a57ff Compare May 12, 2026 17:23

parambole force-pushed the deepseek_v4_compressed_attention branch from 4520166 to 10ca4f6 Compare May 12, 2026 17:23

parambole force-pushed the dsv4-moe-routing-primitives branch from 22a57ff to 32869e5 Compare May 12, 2026 21:12

parambole force-pushed the deepseek_v4_compressed_attention branch from 10ca4f6 to 31a5932 Compare May 12, 2026 21:13

parambole force-pushed the dsv4-moe-routing-primitives branch from 32869e5 to c92f2e0 Compare May 14, 2026 17:51

parambole force-pushed the deepseek_v4_compressed_attention branch from 31a5932 to c98a34e Compare May 14, 2026 17:53

parambole changed the title ~~Implement DeepSeek-V4 Compressed Attention Layers~~ [DeepSeek-V4] Implement Compressed Attention Layers May 14, 2026

parambole added the gemini-review label May 14, 2026

parambole marked this pull request as ready for review May 14, 2026 18:57

parambole requested review from NicoGrande, NuojCheng, RissyRan, abhinavclemson, bvandermoon, gagika, gobbleturk, jesselu-google, jiangjy1982, khatwanimohit, richjames0, shralex, shuningjin and suexu1025 as code owners May 14, 2026 18:57

parambole requested review from A9isha, SurbhiJainUSC, aireenmei, dipannita08, hengtaoguo, igorts-git and vipannalla as code owners May 14, 2026 18:57

parambole assigned entrpn, RissyRan, shuningjin, gagika, snehalv2002, Rohan-Bierneni and bvandermoon May 14, 2026

parambole added gemini-review and removed gemini-review labels May 14, 2026

github-actions Bot reviewed May 14, 2026

View reviewed changes

RissyRan reviewed May 16, 2026

View reviewed changes

parambole added 4 commits May 20, 2026 17:41

feat(attention): update CSA/HCA compressors and Lightning Indexer que…

327a68d

…ry expansion parameters for Trillium parity

Merge dsv4-moe-routing-primitives into deepseek_v4_compressed_attention

e8a7a6f

Merge dsv4-moe-routing-primitives into deepseek_v4_compressed_attention

adb7774

Merge dsv4-moe-routing-primitives into deepseek_v4_compressed_attention

9d79b99

shuningjin reviewed May 26, 2026

View reviewed changes

entrpn approved these changes May 26, 2026

View reviewed changes

-    logits = jnp.einsum("bhsd, bhkd -> bhsk", q, k, precision=self.config.matmul_precision) * self.scaling
+    # Broadcast key/value configurations to all heads using broadcasting in einsum
+    # k and v remain [B, 1, S_kv, D_head]
+    k = kv
+    v = kv
+    # Compute attention logits with head broadcasting: [B, H, S, S_kv]
+    logits = jnp.einsum("bhsd, b1kd -> bhsk", q, k, precision=self.config.matmul_precision) * self.scaling

	attn_output = jnp.einsum("bhsk, bhkd -> bhsd", attn_weights, v, precision=self.config.matmul_precision)
	# Project attention weights onto values with head broadcasting: [B, H, S, D_head]
	attn_output = jnp.einsum("bhsk, b1kd -> bhsd", attn_weights, v, precision=self.config.matmul_precision)

-    scores = jnp.matmul(q, swaped_kv)
+    # swapped_kv: [B, 1, D_idx, W]
+    swapped_kv = jnp.swapaxes(compressed_kv, -1, -2)
+    swapped_kv = jnp.expand_dims(swapped_kv, axis=1)
+    # scores: [B, S, H, W]
+    scores = jnp.matmul(q, swapped_kv)

		return compressed_kv, block_bias


		class DeepSeekV4Indexer(nnx.Module):

	# DeepSeek Sparse Attention (DSA)
	# deepseek3.2 introduces indexer in MLA
	use_indexer: False
	indexer_head_dim: 128
	indexer_n_heads: 64
	indexer_topk: 2048

	attention_type: 'global' # Supported attention_type: global, local_sliding, chunk, mla
	share_kv_projections: False # Note: Not compatible with attention_type='mla'
	attention_bias: False # If True, adds a learnable bias to the query, key, and value projections
	attention_sink: False
	sliding_window_size: 0

	return DenseGeneral(
	in_features_shape=self.convert_dense_general_inputs_shape(inputs_kv_shape),
	out_features_shape=(self.num_kv_heads, self.head_dim),
	axis=-1,
	kernel_init=self.kernel_init,
	kernel_axes=kernel_axes,
	dtype=self.dtype,
	weight_dtype=self.weight_dtype,
	quant=self.quant,
	shard_mode=self.config.shard_mode,
	matmul_precision=self.config.matmul_precision,
	use_bias=self.use_bias_in_projections,
	rngs=self.rngs,
	)

	def _init_projections(self, inputs_q_shape: Tuple, inputs_kv_shape: Tuple) -> None:
	"""Initializes the MLA-specific projections."""
	# Assert required configuration parameters for MLA attention.
	assert (
	self.config.attention_type == AttentionType.MLA.value
	), f"MLA requires MLA attention type {AttentionType.MLA.value}"
	assert self.kv_lora_rank > 0, "KV LoRA rank must be > 0"
	assert self.qk_nope_head_dim > 0, "QK NoPe head dim must be > 0"
	assert self.qk_rope_head_dim > 0, "QK RoPE head dim must be > 0"
	assert self.v_head_dim > 0, "V head dim must be > 0"
	assert self.num_query_heads == self.num_kv_heads, "MLA requires equal number of query and kv heads"
	assert not self.config.fused_qkv, "Fused QKV is not supported for MLA"

	if self.q_lora_rank == 0:
	# Standard Q projection (without LoRA).
	self.query = DenseGeneral(
	in_features_shape=self.config.emb_dim,
	out_features_shape=(self.num_query_heads, self.qk_head_dim),
	axis=-1,
	kernel_init=self.kernel_init,
	kernel_axes=("embed", "q_heads", "kv"),
	dtype=self.dtype,
	weight_dtype=self.weight_dtype,
	quant=self.quant,
	matmul_precision=self.config.matmul_precision,
	shard_mode=self.config.shard_mode,
	rngs=self.rngs,
	)

		# Projections for query extraction and low-rank compression
		self.q_a_proj = nnx.Linear(

	# A large negative mask value is used for masking to ensure that the
	# softmax function assigns an extremely low probability to the masked positions.
	DEFAULT_MASK_VALUE = -0.7 * float(np.finfo(np.dtype("float32")).max)

Conversation

parambole commented May 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Tests

Checklist

Uh oh!

codecov Bot commented May 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

github-actions Bot commented May 14, 2026

Uh oh!

github-actions Bot commented May 14, 2026

Uh oh!

github-actions Bot left a comment

Choose a reason for hiding this comment

🔍 General Feedback

Uh oh!

github-actions Bot May 14, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions Bot May 14, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions Bot May 14, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions Bot May 14, 2026

Choose a reason for hiding this comment

Uh oh!

RissyRan left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

shuningjin left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

entrpn left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

8 participants

parambole commented May 11, 2026 •

edited

Loading

codecov Bot commented May 11, 2026 •

edited

Loading