feat: real cp support with relayout fix for qwen3.5 train/rollout mismatch by Zhichenzzz · Pull Request #885 · radixark/miles

Zhichenzzz · 2026-04-04T23:35:32Z

Summary

This PR addresses #878 , adding real Context Parallelism (CP) support and fixing a train/rollout mismatch bug that was overlooked in PrimeIntellect prime-rl#2080.

Problem

When using hybrid linear attention with CP, there is a silent but critical layout mismatch between training and rollout: flash-linear-attention uses a zig-zag layout for CP, while Megatron uses a packed layout. The approach taken in prime-rl #2080 did not account for this incompatibility, resulting in undetected logprob divergence during training.

Solution

We identified this root cause and implemented proper shard and reshard operations to correctly bridge the zig-zag ↔ packed layout between flash-linear-attention and Megatron under CP, ensuring full consistency between train and rollout.

Results

After 10 training steps, logprob diff across configurations:

Config	logprob_diff
CP=1, EP=4	0.0121
CP=2, EP=4	0.0113
CP=2, EP=8	0.0105

With the shard/reshard fix applied, logprob diff is further reduced to ~0.001. Thanks @guapisolo for the great help!

References

Closes [Feature] Support real CP for Qwen 3.5 #878
Related (flawed approach): PrimeIntellect/prime-rl#2080

- Upgrade flash-linear-attention to 0.4.2 (adds fla.ops.cp module) - Pass cp_context to conv1d for correct boundary token handling - Pass cp_context to chunk_gated_delta_rule for recurrent state CP - Add use_native_cp flag to skip all-gather for DeltaNet layers - Add setup_hybrid_cp() to configure GDN modules for native CP - Add correctness test (torchrun) and e2e test for CP=2/4 - Add CP=2 EP=8 training script Fixes: #878

gemini-code-assist

Code Review

This pull request implements native Context Parallel (CP) for GatedDeltaNet layers in Qwen3.5 models, updating the flash-linear-attention dependency and adding initialization logic for hybrid CP. It also refines Hugging Face configuration loading and validation, particularly for MoE models, and introduces comprehensive E2E and correctness tests. Feedback highlights a critical issue where the CP context construction assumes a batch size of one, which would cause failures in multi-sequence batches, and suggests more robust validation for MoE intermediate sizes by checking moe_intermediate_size instead of skipping the check entirely.

gemini-code-assist · 2026-04-04T23:37:59Z

+    is_moe = hasattr(hf_config, "num_experts") or hasattr(hf_config, "moe_intermediate_size")
+    checks = [
        ("hidden_size", "hidden_size", equal),
        ("num_attention_heads", "num_attention_heads", equal),
        ("num_hidden_layers", "num_layers", equal),
-        ("intermediate_size", "ffn_hidden_size", equal),
        ("tie_word_embeddings", "untie_embeddings_and_output_weights", lambda x, y: not x == y),
        (
            "rms_norm_eps",
            "norm_epsilon" if os.getenv("DEPRECATED_MEGATRON_COMPATIBLE", "0") == "1" else "layernorm_epsilon",
            equal,
        ),
        ("rope_theta", "rotary_base", equal),
-    ]:
+    ]
+    if not is_moe:
+        checks.insert(3, ("intermediate_size", "ffn_hidden_size", equal))


The current implementation skips the validation of the intermediate size for MoE models. For MoE models, ffn_hidden_size should be validated against moe_intermediate_size (if present) or intermediate_size. Hardcoding the skip for all MoE models reduces the robustness of the configuration check.

Suggested change

is_moe = hasattr(hf_config, "num_experts") or hasattr(hf_config, "moe_intermediate_size")

checks = [

("hidden_size", "hidden_size", equal),

("num_attention_heads", "num_attention_heads", equal),

("num_hidden_layers", "num_layers", equal),

("intermediate_size", "ffn_hidden_size", equal),

("tie_word_embeddings", "untie_embeddings_and_output_weights", lambda x, y: not x == y),

(

"rms_norm_eps",

"norm_epsilon" if os.getenv("DEPRECATED_MEGATRON_COMPATIBLE", "0") == "1" else "layernorm_epsilon",

equal,

),

("rope_theta", "rotary_base", equal),

]:

]

if not is_moe:

checks.insert(3, ("intermediate_size", "ffn_hidden_size", equal))

is_moe = hasattr(hf_config, "num_experts") or hasattr(hf_config, "moe_intermediate_size")

intermediate_size_attr = "moe_intermediate_size" if is_moe and hasattr(hf_config, "moe_intermediate_size") else "intermediate_size"

checks = [

("hidden_size", "hidden_size", equal),

("num_attention_heads", "num_attention_heads", equal),

("num_hidden_layers", "num_layers", equal),

(intermediate_size_attr, "ffn_hidden_size", equal),

("tie_word_embeddings", "untie_embeddings_and_output_weights", lambda x, y: not x == y),

(

"rms_norm_eps",

"norm_epsilon" if os.getenv("DEPRECATED_MEGATRON_COMPATIBLE", "0") == "1" else "layernorm_epsilon",

equal,

),

("rope_theta", "rotary_base", equal),

]

References

Model parameters should be retrieved from the model configuration rather than being hardcoded.

gemini-code-assist · 2026-04-04T23:37:59Z

+    def _build_cp_context(self, local_seq_len: int, device: torch.device):
+        """Build fla CP context from the local (sharded) sequence length."""
+        cp_group = getattr(self, "cp_group", None)
+        if cp_group is None or build_cp_context is None:
+            return None
+        global_seq_len = local_seq_len * self.cp_world_size
+        global_cu_seqlens = torch.tensor([0, global_seq_len], dtype=torch.int32, device=device)
+        return build_cp_context(
+            cu_seqlens=global_cu_seqlens,
+            group=cp_group,
+            conv1d_kernel_size=self.conv_kernel_size,
+        )


The _build_cp_context method currently constructs global_cu_seqlens for only a single sequence, which will lead to incorrect behavior or crashes when the batch size is greater than 1. It should be updated to account for the batch size by creating a cu_seqlens tensor that covers all sequences in the batch.

def _build_cp_context(self, batch_size: int, local_seq_len: int, device: torch.device): """Build fla CP context from the local (sharded) sequence length.""" cp_group = getattr(self, "cp_group", None) if cp_group is None or build_cp_context is None: return None cp_world_size = getattr(self, "cp_world_size", 1) global_seq_len = local_seq_len * cp_world_size global_cu_seqlens = torch.arange( 0, (batch_size + 1) * global_seq_len, step=global_seq_len, dtype=torch.int32, device=device ) return build_cp_context( cu_seqlens=global_cu_seqlens, group=cp_group, conv1d_kernel_size=self.conv_kernel_size, )

References

Avoid hardcoding model dimensions; derive them from configuration or input tensor shapes instead.

gemini-code-assist · 2026-04-04T23:37:59Z

    ):
        batch_size, seq_len, _ = hidden_states.shape

+        cp_context = self._build_cp_context(seq_len, hidden_states.device)


The call to _build_cp_context needs to pass the batch_size to correctly initialize the CP context for multi-sequence batches.

Suggested change

cp_context = self._build_cp_context(seq_len, hidden_states.device)

cp_context = self._build_cp_context(batch_size, seq_len, hidden_states.device)

gemini-code-assist · 2026-04-04T23:37:59Z

+    def _build_cp_context(self, local_seq_len: int, device: torch.device):
+        """Build fla CP context from the local (sharded) sequence length."""
+        cp_group = getattr(self, "cp_group", None)
+        if cp_group is None or build_cp_context is None:
+            return None
+        global_seq_len = local_seq_len * self.cp_world_size
+        global_cu_seqlens = torch.tensor([0, global_seq_len], dtype=torch.int32, device=device)
+        return build_cp_context(
+            cu_seqlens=global_cu_seqlens,
+            group=cp_group,
+            conv1d_kernel_size=self.conv_kernel_size,
+        )


The _build_cp_context method ignores the batch size when constructing global_cu_seqlens. This will cause issues during Context Parallel execution if the batch size is greater than 1. The cu_seqlens should be generated based on the batch size.

def _build_cp_context(self, batch_size: int, local_seq_len: int, device: torch.device): """Build fla CP context from the local (sharded) sequence length.""" cp_group = getattr(self, "cp_group", None) if cp_group is None or build_cp_context is None: return None cp_world_size = getattr(self, "cp_world_size", 1) global_seq_len = local_seq_len * cp_world_size global_cu_seqlens = torch.arange( 0, (batch_size + 1) * global_seq_len, step=global_seq_len, dtype=torch.int32, device=device ) return build_cp_context( cu_seqlens=global_cu_seqlens, group=cp_group, conv1d_kernel_size=self.conv_kernel_size, )

References

Avoid hardcoding model dimensions; derive them from configuration or input tensor shapes instead.

gemini-code-assist · 2026-04-04T23:37:59Z

        hidden_states: torch.Tensor,
        cu_seqlens: torch.Tensor = None,
    ):
+        cp_context = self._build_cp_context(hidden_states.shape[1], hidden_states.device)


Update the call to _build_cp_context to include the batch size, ensuring the CP context is correctly built for batches with more than one sequence.

Suggested change

cp_context = self._build_cp_context(hidden_states.shape[1], hidden_states.device)

cp_context = self._build_cp_context(hidden_states.shape[0], hidden_states.shape[1], hidden_states.device)

- test_qwen3_5_cp_correctness.py → tests/e2e/precision/ - test_qwen3_5_mtp_bridge_mapping.py → tests/e2e/megatron/

Zhichenzzz · 2026-04-06T02:30:47Z

        if "rope_theta" in hf_config.rope_parameters:
            hf_config.rope_theta = hf_config.rope_parameters["rope_theta"]

+    is_moe = hasattr(hf_config, "num_experts") or hasattr(hf_config, "moe_intermediate_size")


@guapisolo I added this line to ensure the qwen 3.5 hf_config intermediate_size=5632 not equal with ffn_hidden_size

Let me double check.

I think we can remove this line? This line is not related to your motivation.

Curious that around few weeks before when I integrated qwen3.5, there was no assert errors here. (what about you I believe you also tested it before) is it from transformer version update or sth else?

guapisolo · 2026-04-06T22:53:01Z

        ("rope_theta", "rotary_base", equal),
    ]:
        if hasattr(hf_config, hf_config_name):
+            if is_moe and hf_config_name == "intermediate_size":


On my side this line might be redundant? https://huggingface.co/Qwen/Qwen3.5-35B-A3B/blob/main/config.json . There is no intermediate_size=5632 in qwen3.5

oh i found it... we should update into pip install transformers==5.2.0.

I add a walkaround to avoid this under transformers 4.57.1

Co-authored-by: Zhichenzzz <zczeng@uw.edu>

guapisolo

LGTM. On my side, Qwen-3.5-35b-a3b, cp=2 logprob diff 0.013
Need code owner review.

…match (radixark#885)

…region clusters (#10) * Revert "[BUGFIX] [P2PRDMA] Add rollout post-processing after P2PRDMA weight updates" (radixark#882) * [Fix] fix ci (radixark#894) * Avoid threading for ray getting object (radixark#886) * Add explicit errors for unsupported Megatron profiles (radixark#887) * Add nvfp4 quantizer files (radixark#907) * Bump flash-linear-attention version to 0.4.2 (radixark#892) * [BUGFIX] Invoke "post_process_quantization" by default after weight updating (radixark#890) Co-authored-by: Yueming Yuan <yym022502@gmail.com> * Add heartbeat and id to session server (radixark#866) * fix: adding thin glm5 image to docker build + latest tag sync (radixark#871) * Add consistent hashing routing policy for rollout (radixark#891) Co-authored-by: Yueming Yuan <yueming@Mac.attlocal.net> * [example] add retool v2 example with multi-turn framework interfaces (radixark#654) Co-authored-by: GuanxingLu <gxlu02@gmail.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * Expose rollout-batch-size, n-samples-per-prompt, global-batch-size as CLI args in swe-agent-v2 (radixark#954) Co-authored-by: Shi Dong <shi.dong@radixark.ai> * chore: remove obsolete swe-agent server.py and run-qwen3.sh (radixark#952) Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * Add weight staleness control for fully async rollout (radixark#958) * Fix/pause generation mode (radixark#924) Co-authored-by: Yueming Yuan <yym022502@gmail.com> * [v0.5.10][1] Bump sglang to v0.5.10 (radixark#898) * [v0.5.10][2] Fix apply_chat_template behavior for transformers >=5.0 (radixark#926) Co-authored-by: guapisolo <guapisolo@gmail.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * [v0.5.10][3] Fix processor return_tensors duplicate kwarg for transformers >=5.0 (radixark#927) Co-authored-by: guapisolo <guapisolo@gmail.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * [v0.5.10][4] Fix _no_split_modules set not subscriptable in transformers >=5.0 (radixark#931) * [v0.5.10][5] Disable piecewise cuda graph to avoid NVLS oom (radixark#935) * [v0.5.10][6][FSDP] fix outdated weight update logic in FSDP (radixark#948) Co-authored-by: guapisolo <guapisolo@gmail.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> Co-authored-by: maocheng23 <35615230+maocheng23@users.noreply.github.com> Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com> * [v0.5.10][7][FSDP] move FSDP to experimental and disable by default (radixark#961) * Add skiplist and more robust calculation on val (radixark#965) * [fix] tiny fix debug rollout only in weight version check (radixark#967) * feat: real cp support with relayout fix for qwen3.5 train/rollout mismatch (radixark#885) * [AMD] Upgrade to sglv0.5.10 (radixark#973) * switch model to actor (radixark#756) * [fix] support general logic to bypass fp32 downcast and fix qwen35 A_log dtype (radixark#975) Co-authored-by: yueming-yuan <yym022502@gmail.com> * fix: populate prefix_cache_info in OpenAI/session rollout path (radixark#960) * Remove prepare_harbor_tasks.py; use harbor-private adapters (radixark#982) * [fix] Skip flush_cache in in_place mode and add fully async example (radixark#974) Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * GLM47 full cmd for async and sync reasoning (radixark#986) Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: handle non-tool appended messages in TITO incremental tokenization (radixark#949) Co-authored-by: Yanbin Jiang <jybsuper@gmail.com> * [docker] Add sgl-model-gateway install and download .tar.gz assets (radixark#895) * [ci] fix hf rate limit error by caching tokenizer loading (radixark#1014) Co-authored-by: maocheng23 <35615230+maocheng23@users.noreply.github.com> * Use load_generate_function in legacy sglang_rollout path (radixark#1016) * Update CODEOWNERS to add new reviewers (radixark#1021) * Support moe lora for gpt-oss (radixark#798) Co-authored-by: Ethan (Yusheng) Su <yushengsu.thu@gmail.com> * [fix] restore expert_bias to fp32 before bridge weight export (radixark#811) * [chore] drop legacy transformers upgrade pin for glm47-flash and qwen35 (radixark#1018) * [fix] Enforce param dtype before wrap ddp (radixark#992) Co-authored-by: Zhichen Zeng <zczeng@uw.edu> * [upgrade] update Megatron-Bridge source and LoRA CI to megatron e2e tests and (radixark#1023) * [CI] Drop --use-miles-router from R3 tests and add r3 comparasion test between sgl & miles router (radixark#1015) * wandb: raise init_timeout, add retry wrapper, fix shared-mode init for cross-region clusters In online + shared mode, both `init_wandb_primary` and `init_wandb_secondary` make HTTPS round-trips to wandb cloud (login + run create/attach). On high-latency cross-region clusters (e.g. Abu Dhabi MBZUAI ↔ wandb-cloud US-West) with concurrent actor bursts, a single round-trip can exceed the wandb SDK's 90s default `init_timeout` — tearing down the whole run with a silent handshake abort. Observed on RL360 job 1564420, which forced `WANDB_MODE=offline` as a global default ever since (see https://github.com/LLM360/RL360/issues/87). The issue's original diagnosis assumed a local primary↔secondary socket handshake race. That's not how shared mode works — per wandb's own feature PR (wandb/wandb#6882), each writer spawns an independent wandb-core that talks to the cloud directly; aggregation is server-side by run_id. No local socket exists. The failure mode is pure network/latency, not a local readiness race. Changes ------- - Bump `init_timeout` to 300s for primary and secondary Settings. Configurable via `WANDB_INIT_TIMEOUT_SECS` env var for tuning. - Wrap both init paths in a bounded exponential-backoff retry (`_wandb_init_with_retry`) that re-attempts on wandb.errors.CommError and wandb.errors.UsageError. 3 attempts with 5→10→20s backoff by default, tunable via `WANDB_INIT_RETRY_ATTEMPTS` / `WANDB_INIT_RETRY_BACKOFF_SECS`. - Add `x_label` tagging per wandb distributed-training docs: primary gets `rank_<rank>_primary`, secondaries get `rank_<rank>_secondary`. Enables per-rank console-log filtering in the wandb UI. - Drop `reinit=True` from secondary init_kwargs. Shared mode natively supports concurrent writers on a single run; `reinit=True` triggered stale-state warnings on secondary actors without functional benefit. Followups this change enables ----------------------------- - `WANDB_MODE=offline` can be removed from scale.yaml's extra_env default once a pilot run confirms online mode boots cleanly. - The tmux-based `~/bin/wandb-sync-rl360.sh` workaround on David's M2 account becomes obsolete (no more offline-only default). - Near-realtime wandb dashboards replace the ~2-minute-lag offline sync; per-rank system metrics via x_label filtering. --------- Co-authored-by: JD <jaedon.guo@gmail.com> Co-authored-by: Ethan (Yusheng) Su <yushengsu.thu@gmail.com> Co-authored-by: fzyzcjy <5236035+fzyzcjy@users.noreply.github.com> Co-authored-by: Ziang Li <ziangli@umich.edu> Co-authored-by: Zhichen Zeng <zczeng@uw.edu> Co-authored-by: JensenFire <xinji1@microsoft.com> Co-authored-by: Yueming Yuan <yym022502@gmail.com> Co-authored-by: maocheng23 <35615230+maocheng23@users.noreply.github.com> Co-authored-by: Douglas Yang <douglasyang88@gmail.com> Co-authored-by: Yueming Yuan <yueming@Mac.attlocal.net> Co-authored-by: Huapeng Zhou <73010314+PopSoda2002@users.noreply.github.com> Co-authored-by: GuanxingLu <gxlu02@gmail.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> Co-authored-by: Shi-Dong <Shi-Dong@users.noreply.github.com> Co-authored-by: Shi Dong <shi.dong@radixark.ai> Co-authored-by: Jiajun Li <48857426+guapisolo@users.noreply.github.com> Co-authored-by: guapisolo <guapisolo@gmail.com> Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com> Co-authored-by: Yuzhen Zhou <82826991+zyzshishui@users.noreply.github.com> Co-authored-by: Yanbin Jiang <jybsuper@gmail.com> Co-authored-by: Ying Sheng <sqy1415@gmail.com> Co-authored-by: Yisheng Gong <yishenggong9437@gmail.com>

gemini-code-assist Bot reviewed Apr 4, 2026

View reviewed changes

alexnails force-pushed the feat/qwen35_cp branch from a7083b5 to f1b2890 Compare April 5, 2026 21:09

del logs

e42a5d2

Zhichenzzz force-pushed the feat/qwen35_cp branch from f1b2890 to e42a5d2 Compare April 5, 2026 21:16

Zhichenzzz added 5 commits April 5, 2026 21:24

chore: move qwen3.5 tests to e2e subdirectories

5d74611

- test_qwen3_5_cp_correctness.py → tests/e2e/precision/ - test_qwen3_5_mtp_bridge_mapping.py → tests/e2e/megatron/

refactor: rename use_native_cp to hybrid_cp

b99ac8f

revert: remove unnecessary is_moe check in hf_validate_args

302ccfc

revert: restore arguments.py to main, no changes needed

b6b4a1f

fix: skip intermediate_size check for MoE models in hf_validate_args

18699a3

Zhichenzzz force-pushed the feat/qwen35_cp branch from 2ebcf9e to 18699a3 Compare April 5, 2026 22:04

Zhichenzzz marked this pull request as ready for review April 5, 2026 22:09

Zhichenzzz requested review from fzyzcjy, guapisolo, maocheng23, yueming-yuan and yushengsu-thu as code owners April 5, 2026 22:09

Zhichenzzz commented Apr 6, 2026

View reviewed changes

guapisolo reviewed Apr 6, 2026

View reviewed changes

Zhichenzzz added 3 commits April 6, 2026 23:25

del assert check

7af979b

script for transformer version 5.2.0

c5f194d

fix: apply black formatting to test_qwen3_5_35B_A3B_cp.py

89d6dcd

Zhichenzzz requested a review from guapisolo April 7, 2026 02:58

guapisolo and others added 3 commits April 7, 2026 16:12

[Fix] try fix feat/qwen35 cp by upd zigzag (#957)

ea7fafa

Co-authored-by: Zhichenzzz <zczeng@uw.edu>

fix: zigzag-packed_shard relayout

463a6ca

tiny fix

dca77e9

guapisolo approved these changes Apr 9, 2026

View reviewed changes

Zhichenzzz mentioned this pull request Apr 9, 2026

[Feature] Support real CP for Qwen 3.5 #878

Closed

yueming-yuan reviewed Apr 10, 2026

View reviewed changes

Comment thread scripts/run-qwen3.5-35B-A3B-mtp-cp2-ep8.sh Outdated

yueming-yuan reviewed Apr 10, 2026

View reviewed changes

Comment thread miles/backends/training_utils/cp_utils.py Outdated

yueming-yuan reviewed Apr 10, 2026

View reviewed changes

Comment thread miles/backends/megatron_utils/actor.py Outdated

Zhichenzzz added 4 commits April 10, 2026 23:30

fix nits

e304ef7

reformat

4471d46

minor format

2842d57

Merge branch 'main' into feat/qwen35_cp

7314c5b

Zhichenzzz requested a review from yueming-yuan April 11, 2026 00:11

Zhichenzzz changed the title ~~feat: hybrid cp for qwen3.5 rl~~ feat: real cp support with zig-zag/packed layout fix for qwen3.5 train/rollout mismatch Apr 11, 2026

Zhichenzzz changed the title ~~feat: real cp support with zig-zag/packed layout fix for qwen3.5 train/rollout mismatch~~ feat: real cp support with relayout fix for qwen3.5 train/rollout mismatch Apr 11, 2026

yueming-yuan approved these changes Apr 11, 2026

View reviewed changes

Zhichenzzz merged commit eb294e3 into main Apr 11, 2026
17 checks passed

Zhichenzzz deleted the feat/qwen35_cp branch April 11, 2026 03:06

Zhichenzzz mentioned this pull request Apr 14, 2026

[Question] Any plans to support CP for GDN layers? THUDM/slime#1744

Open

3 tasks

GuanxingLu pushed a commit to GuanxingLu/miles that referenced this pull request Apr 21, 2026

feat: real cp support with relayout fix for qwen3.5 train/rollout mis…

fa5fdcc

…match (radixark#885)

	cp_context = self._build_cp_context(seq_len, hidden_states.device)
	cp_context = self._build_cp_context(batch_size, seq_len, hidden_states.device)

	cp_context = self._build_cp_context(hidden_states.shape[1], hidden_states.device)
	cp_context = self._build_cp_context(hidden_states.shape[0], hidden_states.shape[1], hidden_states.device)

Conversation

Zhichenzzz commented Apr 4, 2026 • edited by guapisolo Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Problem

Solution

Results

References

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist Bot Apr 4, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Apr 4, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Apr 4, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Apr 4, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Apr 4, 2026

Choose a reason for hiding this comment

Uh oh!

Zhichenzzz Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

guapisolo Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

guapisolo Apr 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Zhichenzzz Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

guapisolo Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

Zhichenzzz Apr 7, 2026

Choose a reason for hiding this comment

Uh oh!

guapisolo Apr 9, 2026

Choose a reason for hiding this comment

Uh oh!

guapisolo left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Zhichenzzz commented Apr 4, 2026 •

edited by guapisolo

Loading

guapisolo Apr 6, 2026 •

edited

Loading

guapisolo left a comment •

edited

Loading