Skip to content

ckpt: support GLM-4.7-Flash test on ROCm (MI35x) #1620

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

ckpt: support GLM-4.7-Flash test on ROCm (MI35x) #1620

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
20 changes: 15 additions & 5 deletions tests/e2e/ckpt/test_glm47_flash_ckpt.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,6 @@
import os

import torch
from tests.ci.ci_register import register_cuda_ci

import miles.utils.external_utils.command_utils as U
Expand All @@ -10,7 +11,10 @@
ENABLE_EVAL = 0
USE_DEEPEP = 0

TIGHT_HOST_MEMORY = bool(int(os.environ.get("MILES_TEST_TIGHT_HOST_MEMORY", "0")))
ENABLE_EVAL = bool(int(os.environ.get("MILES_TEST_ENABLE_EVAL", "1")))
TIGHT_HOST_MEMORY = bool(int(os.environ.get("MILES_TEST_TIGHT_HOST_MEMORY", "1")))
USE_DEEPEP = bool(int(os.environ.get("MILES_TEST_USE_DEEPEP", "0")))
Comment on lines +14 to +16

@gemini-code-assist gemini-code-assist Bot Jul 10, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

medium

The variables ENABLE_EVAL and USE_DEEPEP are redefined here using environment variables, but their initial static assignments (ENABLE_EVAL = 0 and USE_DEEPEP = 0 on lines 11 and 12) were not removed. This results in redundant assignments and dead code. Please remove the initial assignments on lines 11 and 12.

IS_ROCM = torch.version.hip is not None

MODEL_NAME = "GLM-4.7-Flash"
MODEL_TYPE = "glm4.7-flash"
Expand Down Expand Up @@ -141,6 +145,8 @@ def execute(mode: str = "", ckpt_step: int | None = None):
ci_args += "--ci-save-model-hash "
if mode == "load":
ci_args += "--ci-check-model-hash "
if IS_ROCM:
ci_args += "--ci-disable-logprobs-checker "

misc_args = (
"--attention-dropout 0.0 "
Expand Down Expand Up @@ -172,14 +178,18 @@ def execute(mode: str = "", ckpt_step: int | None = None):
f"{misc_args} "
)

extra_env_vars = {
"MILES_EXPERIMENTAL_ROLLOUT_REFACTOR": "1",
"MILES_TEST_R3_THRESHOLD": "1.0",
}
if IS_ROCM:
extra_env_vars["SGLANG_USE_AITER"] = "0"

U.execute_train(
train_args=train_args,
num_gpus_per_node=NUM_GPUS,
megatron_model_type=MODEL_TYPE,
extra_env_vars={
"MILES_EXPERIMENTAL_ROLLOUT_REFACTOR": "1",
"MILES_TEST_R3_THRESHOLD": "1.0",
},
extra_env_vars=extra_env_vars,
)


Expand Down