sglang-0.5.2-py3-none-any.whl
ERROR: HTTP error 404 while getting http://raw.githubusercontent.com/inclusionAI/Ring-V2/blob/main/hybrid_linear/whls/sglang-0.5.2-py3-none-any.whl
is there new version or was this already merged? How to get the sglang?
@festr2 You could download the whl file from https://github.com/inclusionAI/Ring-V2/tree/main/hybrid_linear
ERROR: HTTP error 404 while getting http://raw.githubusercontent.com/inclusionAI/Ring-V2/blob/main/hybrid_linear/whls/sglang-0.5.2-py3-none-any.whl
is there new version or was this already merged? How to get the sglang?
You can try installing it by following command:
pip install https://raw.githubusercontent.com/inclusionAI/Ring-V2/refs/heads/main/hybrid_linear/whls/sglang-0.5.2-py3-none-any.whl --no-deps --force-reinstall
ERROR: HTTP error 404 while getting http://raw.githubusercontent.com/inclusionAI/Ring-V2/blob/main/hybrid_linear/whls/sglang-0.5.2-py3-none-any.whl
is there new version or was this already merged? How to get the sglang?
you can also find our PR here: https://github.com/sgl-project/sglang/pull/10917
pip install https://raw.githubusercontent.com/inclusionAI/Ring-V2/refs/heads/main/hybrid_linear/whls/sglang-0.5.2-py3-none-any.whl --no-deps --force-reinstall
python -m sglang.launch_server --model-path /mnt/Ring-flash-linear-2.0 --trust-remote-code --tp-size 2 --json-model-override-args "{"linear_backend": "seg_la"}"
File "/usr/local/lib/python3.12/dist-packages/sglang/srt/model_loader/utils.py", line 68, in resolve_transformers_arch
raise ValueError(
ValueError: BailingMoeLinearV2ForCausalLM has no SGlang implementation and the Transformers implementation is not compatible with SGLang.
any hints?
Can you provide more ENV details? It seems like the installation our compiled sglang is still not rightly been installed.
pip install https://raw.githubusercontent.com/inclusionAI/Ring-V2/refs/heads/main/hybrid_linear/whls/sglang-0.5.2-py3-none-any.whl --no-deps --force-reinstall
python -m sglang.launch_server --model-path /mnt/Ring-flash-linear-2.0 --trust-remote-code --tp-size 2 --json-model-override-args "{"linear_backend": "seg_la"}"
File "/usr/local/lib/python3.12/dist-packages/sglang/srt/model_loader/utils.py", line 68, in resolve_transformers_arch
raise ValueError(
ValueError: BailingMoeLinearV2ForCausalLM has no SGlang implementation and the Transformers implementation is not compatible with SGLang.any hints?
This might be caused by an error when SGLang imports bailing_moe_linear.py. Could you please check the versions of the other packages? We are using:
torch: 2.8.0
Python: 3.10.12
transformers: 4.56.1
Please make sure your environment matches these versions, and let us know if the issue persists.
pip install https://raw.githubusercontent.com/inclusionAI/Ring-V2/refs/heads/main/hybrid_linear/whls/sglang-0.5.2-py3-none-any.whl --no-deps --force-reinstall
python -m sglang.launch_server --model-path /mnt/Ring-flash-linear-2.0 --trust-remote-code --tp-size 2 --json-model-override-args "{"linear_backend": "seg_la"}"
File "/usr/local/lib/python3.12/dist-packages/sglang/srt/model_loader/utils.py", line 68, in resolve_transformers_arch
raise ValueError(
ValueError: BailingMoeLinearV2ForCausalLM has no SGlang implementation and the Transformers implementation is not compatible with SGLang.any hints?
This might be caused by an error when SGLang imports bailing_moe_linear.py. Could you please check the versions of the other packages? We are using:
torch: 2.8.0
Python: 3.10.12
transformers: 4.56.1
Please make sure your environment matches these versions, and let us know if the issue persists.
I have tried to compile the latest sglang from source including your PRs and I have the same issue like with your whl.
File "/mnt/sglang/python/sglang/srt/model_loader/utils.py", line 68, in resolve_transformers_arch
raise ValueError(
ValueError: BailingMoeLinearV2ForCausalLM has no SGlang implementation and the Transformers implementation is not compatible with SGLang.
I tried transformers 4.56.1 and also 4.56.2 (same issue)
pytorch-triton 3.4.0+gitf7888497
torch 2.9.0.dev20250901+cu129
Python 3.12.11
pip install https://raw.githubusercontent.com/inclusionAI/Ring-V2/refs/heads/main/hybrid_linear/whls/sglang-0.5.2-py3-none-any.whl --no-deps --force-reinstall
python -m sglang.launch_server --model-path /mnt/Ring-flash-linear-2.0 --trust-remote-code --tp-size 2 --json-model-override-args "{"linear_backend": "seg_la"}"
File "/usr/local/lib/python3.12/dist-packages/sglang/srt/model_loader/utils.py", line 68, in resolve_transformers_arch
raise ValueError(
ValueError: BailingMoeLinearV2ForCausalLM has no SGlang implementation and the Transformers implementation is not compatible with SGLang.any hints?
This might be caused by an error when SGLang imports bailing_moe_linear.py. Could you please check the versions of the other packages? We are using:
torch: 2.8.0
Python: 3.10.12
transformers: 4.56.1
Please make sure your environment matches these versions, and let us know if the issue persists.I have tried to compile the latest sglang from source including your PRs and I have the same issue like with your whl.
File "/mnt/sglang/python/sglang/srt/model_loader/utils.py", line 68, in resolve_transformers_arch
raise ValueError(
ValueError: BailingMoeLinearV2ForCausalLM has no SGlang implementation and the Transformers implementation is not compatible with SGLang.I tried transformers 4.56.1 and also 4.56.2 (same issue)
pytorch-triton 3.4.0+gitf7888497
torch 2.9.0.dev20250901+cu129
Python 3.12.11
You can install the following packages first.pip install sglang==0.5.2 sgl-kernel==0.3.9.post2 vllm==0.10.2 torch==2.8.0 torchvision==0.23.0 torchao
Then install the sglang we adapted.pip install https://raw.githubusercontent.com/inclusionAI/Ring-V2/refs/heads/main/hybrid_linear/whls/sglang-0.5.2-py3-none-any.whl --no-deps --force-reinstall
pip install https://raw.githubusercontent.com/inclusionAI/Ring-V2/refs/heads/main/hybrid_linear/whls/sglang-0.5.2-py3-none-any.whl --no-deps --force-reinstall
python -m sglang.launch_server --model-path /mnt/Ring-flash-linear-2.0 --trust-remote-code --tp-size 2 --json-model-override-args "{"linear_backend": "seg_la"}"
File "/usr/local/lib/python3.12/dist-packages/sglang/srt/model_loader/utils.py", line 68, in resolve_transformers_arch
raise ValueError(
ValueError: BailingMoeLinearV2ForCausalLM has no SGlang implementation and the Transformers implementation is not compatible with SGLang.any hints?
This might be caused by an error when SGLang imports bailing_moe_linear.py. Could you please check the versions of the other packages? We are using:
torch: 2.8.0
Python: 3.10.12
transformers: 4.56.1
Please make sure your environment matches these versions, and let us know if the issue persists.I have tried to compile the latest sglang from source including your PRs and I have the same issue like with your whl.
File "/mnt/sglang/python/sglang/srt/model_loader/utils.py", line 68, in resolve_transformers_arch
raise ValueError(
ValueError: BailingMoeLinearV2ForCausalLM has no SGlang implementation and the Transformers implementation is not compatible with SGLang.I tried transformers 4.56.1 and also 4.56.2 (same issue)
pytorch-triton 3.4.0+gitf7888497
torch 2.9.0.dev20250901+cu129
Python 3.12.11
Could you please check whether the file sglang/srt/models/bailing_moe_linear.py exists in your SGLang installation directory?
If it does exist, please try running the following command in a Python console:
from sglang.srt.models.bailing_moe_linear import BailingMoeLinearV2ForCausalLM
Let us know if any error occurs—we’d appreciate your feedback so we can assist further.
If it does exist, please try running the following command in a Python console:
from sglang.srt.models.bailing_moe_linear import BailingMoeLinearV2ForCausalLM
Let us know if any error occurs—we’d appreciate your feedback so we can assist further.
This file exists:
Python 3.12.11 (main, Jun 4 2025, 08:56:18) [GCC 11.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
from sglang.srt.models.bailing_moe_linear import BailingMoeLinearV2ForCausalLM
Traceback (most recent call last):
File "", line 1, in
File "/mnt/sglang/python/sglang/srt/models/bailing_moe_linear.py", line 10, in
from transformers import PretrainedConfig
File "/mnt/sglang/python/sglang/srt/models/transformers.py", line 24, in
from transformers import AutoModel, PretrainedConfig, PreTrainedModel
ImportError: cannot import name 'AutoModel' from partially initialized module 'transformers' (most likely due to a circular import) (/mnt/sglang/python/sglang/srt/models/transformers.py)
If it does exist, please try running the following command in a Python console:
from sglang.srt.models.bailing_moe_linear import BailingMoeLinearV2ForCausalLM
Let us know if any error occurs—we’d appreciate your feedback so we can assist further.This file exists:
Python 3.12.11 (main, Jun 4 2025, 08:56:18) [GCC 11.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.from sglang.srt.models.bailing_moe_linear import BailingMoeLinearV2ForCausalLM
Traceback (most recent call last):
File "", line 1, in
File "/mnt/sglang/python/sglang/srt/models/bailing_moe_linear.py", line 10, in
from transformers import PretrainedConfig
File "/mnt/sglang/python/sglang/srt/models/transformers.py", line 24, in
from transformers import AutoModel, PretrainedConfig, PreTrainedModel
ImportError: cannot import name 'AutoModel' from partially initialized module 'transformers' (most likely due to a circular import) (/mnt/sglang/python/sglang/srt/models/transformers.py)
This error appears to be caused by an incorrect or incomplete installation of transformers.
You can try uninstalling and reinstalling the correct version using the following commands:
pip uninstall transformers
pip install transformers==4.56.1
I did this and still not working. I gave up. Docker image would be nice