repo stringclasses 147 values | number int64 1 172k | title stringlengths 2 476 | body stringlengths 0 5k | url stringlengths 39 70 | state stringclasses 2 values | labels listlengths 0 9 | created_at timestamp[ns, tz=UTC]date 2017-01-18 18:50:08 2026-01-06 07:33:18 | updated_at timestamp[ns, tz=UTC]date 2017-01-18 19:20:07 2026-01-06 08:03:39 | comments int64 0 58 ⌀ | user stringlengths 2 28 |
|---|---|---|---|---|---|---|---|---|---|---|
vllm-project/vllm | 29,865 | [Doc]: | ### 📚 The doc issue
# Installation des bibliothèques XAI
!pip install shap
!pip install lime
!pip install alibi
!pip install interpret
!pip install dalex
!pip install eli5
### Suggest a potential alternative/fix
# Installation des bibliothèques XAI
!pip install shap
!pip install lime
!pip install alibi
!pip install interpret
!pip install dalex
!pip install eli5
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29865 | closed | [
"documentation"
] | 2025-12-02T10:43:01Z | 2025-12-02T10:50:00Z | 0 | hassaballahmahamatahmat5-cpu |
vllm-project/vllm | 29,864 | [Usage]: I am unable to run the GLM-4.5-Air-REAP-82B-A12B-nvfp4 model on an RTX 5090. | ### Your current environment
I am unable to run the GLM-4.5-Air-REAP-82B-A12B-nvfp4 model on an RTX 5090.
```text
Collecting environment information...
==============================
System Info
==============================
OS : Ubuntu 22.04.5 LTS (x86_64)
GCC version : (Ubuntu 12.3.0-1ubuntu1~22.04.2) 12.3.0
Clang version : Could not collect
CMake version : version 4.2.0
Libc version : glibc-2.35
==============================
PyTorch Info
==============================
PyTorch version : 2.10.0.dev20251124+cu128
Is debug build : False
CUDA used to build PyTorch : 12.8
ROCM used to build PyTorch : N/A
==============================
Python Environment
==============================
Python version : 3.10.12 (main, Nov 4 2025, 08:48:33) [GCC 11.4.0] (64-bit runtime)
Python platform : Linux-6.8.0-87-generic-x86_64-with-glibc2.35
==============================
CUDA / GPU Info
==============================
Is CUDA available : True
CUDA runtime version : 12.8.93
CUDA_MODULE_LOADING set to :
GPU models and configuration :
GPU 0: NVIDIA GeForce RTX 5090
GPU 1: NVIDIA GeForce RTX 5090
GPU 2: NVIDIA GeForce RTX 5090
GPU 3: NVIDIA GeForce RTX 5090
Nvidia driver version : 570.172.08
cuDNN version : Could not collect
HIP runtime version : N/A
MIOpen runtime version : N/A
Is XNNPACK available : True
==============================
CPU Info
==============================
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 46 bits physical, 57 bits virtual
Byte Order: Little Endian
CPU(s): 112
On-line CPU(s) list: 0-111
Vendor ID: GenuineIntel
Model name: Intel(R) Xeon(R) Gold 6330 CPU @ 2.00GHz
CPU family: 6
Model: 106
Thread(s) per core: 2
Core(s) per socket: 28
Socket(s): 2
Stepping: 6
CPU max MHz: 3100.0000
CPU min MHz: 800.0000
BogoMIPS: 4000.00
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 intel_ppin ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect wbnoinvd dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req vnmi avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid fsrm md_clear pconfig flush_l1d arch_capabilities
Virtualization: VT-x
L1d cache: 2.6 MiB (56 instances)
L1i cache: 1.8 MiB (56 instances)
L2 cache: 70 MiB (56 instances)
L3 cache: 84 MiB (2 instances)
NUMA node(s): 2
NUMA node0 CPU(s): 0-27,56-83
NUMA node1 CPU(s): 28-55,84-111
Vulnerability Gather data sampling: Mitigation; Microcode
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Enhanced / Automatic IBRS; IBPB conditional; RSB filling; PBRSB-eIBRS SW sequence; BHI SW loop, KVM SW loop
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected
| https://github.com/vllm-project/vllm/issues/29864 | open | [
"usage"
] | 2025-12-02T10:13:31Z | 2025-12-05T17:06:30Z | 2 | east612-ai |
huggingface/diffusers | 12,772 | How to convert diffusers model to wan2.2 format | I see convert_wan_to_diffusers.py in diffusers repo, but no convert_diffusers_to_wan.py. Do you have plan to upload a convert scripts?
| https://github.com/huggingface/diffusers/issues/12772 | open | [] | 2025-12-02T09:19:29Z | 2025-12-02T09:19:29Z | null | wikiwen |
huggingface/diffusers | 12,764 | When will the img2img pipeline of FLUX.2-dev be released? | I see that the current version(0.36.0-dev) only updated the text-to-image pipeline for Flux2. We are looking forward to the update of the image-to-image pipeline!
| https://github.com/huggingface/diffusers/issues/12764 | open | [] | 2025-12-01T11:25:35Z | 2025-12-01T11:41:56Z | 1 | guanxyu |
huggingface/smolagents | 1,890 | Question: how to use sever-side tools provided by Google Gemini or OpenAI GPT? | Gemini has some server-side tools like google_search (https://ai.google.dev/gemini-api/docs/google-search) or google_map. OpenAI also has server-side tools like web_search. Does Smolagents support using such server-side tools from agents? If so, how? | https://github.com/huggingface/smolagents/issues/1890 | open | [] | 2025-12-01T05:16:01Z | 2025-12-23T10:49:45Z | null | victorx-deckard |
huggingface/agents-course | 623 | Message: Submission received, but no valid/matching task IDs were found in the 1 answers provided. Score did not improve previous record, leaderboard not updated. | I am correctly downloading the GAIA 2023 Level 1 validation dataset using snapshot_download and load_dataset. This submission is for Unit 4 Agent Course.
data_dir = snapshot_download(
repo_id="gaia-benchmark/GAIA",
repo_type="dataset"
)
dataset = load_dataset(data_dir, "2023_level1", split="validation")
subset = dataset.select(range(20))
for item in subset:
task_id = item.get("task_id")
question_text = item.get("Question")
file_name = item.get("file_name")
I experience failures when trying to run the first 20 questions i received only 5 task ids are valid.. When I specifically tried to isolate and run the task ID '935e2cff-ae78-4218-b3f5-115589b19dae' using the filtering method, the evaluation system reported.
<img width="1388" height="668" alt="Image" src="https://github.com/user-attachments/assets/f4e9fe1b-8608-4ab4-84cc-1b196a601694" />
'Submission received, but no valid/matching task IDs were found in the 1 answers provided.' This occurred even though I was confident the answer was correct | https://github.com/huggingface/agents-course/issues/623 | open | [
"question"
] | 2025-12-01T02:09:21Z | 2025-12-01T02:09:21Z | null | ShwetaBorole |
huggingface/tokenizers | 1,902 | Guide: Compiling `tokenizers` on Android/Termux | Hello Hugging Face team and fellow developers,
This is a guide for anyone trying to install `tokenizers` (or packages that depend on it, like `transformers` or `docling`) on an Android device using [Termux](https://termux.dev/). Currently, there are no other issues mentioning Termux, so hopefully, this guide can help others.
### The Problem
When running `pip install tokenizers` in a standard Termux environment, the installation fails during the compilation of a C++ dependency with an error similar to this:
```
error: use of undeclared identifier 'pthread_cond_clockwait'
```
This happens because the build system is targeting an Android API level where this function is not available in the C library headers.
### The Solution
The solution is to force the compilation from source and pass specific flags to the C++ compiler to set the correct Android API level and link the required libraries.
Here is a step-by-step guide:
#### Step 1: Install Build Dependencies
You will need the Rust toolchain and other build essentials. You can install them in Termux using `pkg`:
```bash
pkg update && pkg install rust clang make maturin
```
#### Step 2: Find Your Android API Level
The fix requires telling the compiler which Android API level you are using. You can get this number by running the following command in your Termux shell:
```bash
getprop ro.build.version.sdk
```
This will return a number, for example `29`, `30`, `33`, etc. This function (`pthread_cond_clockwait`) was introduced in API level 21, so your device's level should be higher than that.
#### Step 3: Compile and Install `tokenizers`
Now, you can install the package using `pip`. The command below will automatically use the API level from the previous step.
```bash
# This command automatically gets your API level and uses it to compile tokenizers
ANDROID_API_LEVEL=$(getprop ro.build.version.sdk)
CXXFLAGS="-lpthread -D__ANDROID_API__=${ANDROID_API_LEVEL}" pip install tokenizers --no-binary :all:
```
After this, `pip install tokenizers` (and packages that depend on it) should succeed.
#### Explanation of the Flags:
* `CXXFLAGS="..."`: This sets environment variables to pass flags to the C++ compiler.
* `-lpthread`: This flag explicitly tells the linker to link against the POSIX threads library.
* `-D__ANDROID_API__=${ANDROID_API_LEVEL}`: This is the critical part. It defines a macro that tells the C++ headers to expose functions available for your specific Android version, making `pthread_cond_clockwait` visible to the compiler.
* `--no-binary :all:`: This forces `pip` to ignore pre-compiled wheels and build the package from the source code, which is necessary for the flags to be applied.
Hope this helps other developers working in the Termux environment! | https://github.com/huggingface/tokenizers/issues/1902 | open | [] | 2025-12-01T00:46:42Z | 2025-12-01T00:46:42Z | 0 | Manamama-Gemini-Cloud-AI-01 |
vllm-project/vllm | 29,747 | [Bug]: --scheduling-policy=priority & n>1 crashes engine | ### Your current environment
<details>
<summary>The output of <code>python collect_env.py</code></summary>
```text
Your output of `python collect_env.py` here
```
</details>
### 🐛 Describe the bug
When running with priority scheduling, e.g.:
```bash
vllm serve Qwen/Qwen3-0.6B --scheduling-policy=priority
```
and using `n` > 1 in the request, like:
```python
from openai import OpenAI
client = OpenAI(base_url="http://localhost:8000/v1", api_key="dummy")
res = client.chat.completions.create(
model=client.models.list().data[0].id,
messages=[{"role": "user", "content": "What is the meaning of life?"}],
n=2
)
print(res)
```
vllm crashes with:
```python
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] EngineCore encountered a fatal error.
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] Traceback (most recent call last):
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] File "/home/user/code/debug/.venv/lib/python3.10/site-packages/vllm/v1/engine/core.py", line 835, in run_engine_core
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] engine_core.run_busy_loop()
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] File "/home/user/code/debug/.venv/lib/python3.10/site-packages/vllm/v1/engine/core.py", line 860, in run_busy_loop
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] self._process_input_queue()
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] File "/home/user/code/debug/.venv/lib/python3.10/site-packages/vllm/v1/engine/core.py", line 885, in _process_input_queue
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] self._handle_client_request(*req)
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] File "/home/user/code/debug/.venv/lib/python3.10/site-packages/vllm/v1/engine/core.py", line 907, in _handle_client_request
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] self.add_request(req, request_wave)
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] File "/home/user/code/debug/.venv/lib/python3.10/site-packages/vllm/v1/engine/core.py", line 291, in add_request
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] self.scheduler.add_request(request)
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] File "/home/user/code/debug/.venv/lib/python3.10/site-packages/vllm/v1/core/sched/scheduler.py", line 1242, in add_request
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] self.waiting.add_request(request)
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] File "/home/user/code/debug/.venv/lib/python3.10/site-packages/vllm/v1/core/sched/request_queue.py", line 150, in add_request
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] heapq.heappush(self._heap, (request.priority, request.arrival_time, request))
(EngineCore_DP0 pid=207394) ERROR 11-30 15:14:29 [core.py:844] TypeError: '<' not supported between instances of 'Request' and 'Request'
(EngineCore_DP0 pid=207394) Process EngineCore_DP0:
(APIServer pid=207278) ERROR 11-30 15:14:29 [async_llm.py:525] AsyncLLM output_handler failed.
(APIServer pid=207278) ERROR 11-30 15:14:29 [async_llm.py:525] Traceback (most recent call last):
(APIServer pid=207278) ERROR 11-30 15:14:29 [async_llm.py:525] File "/home/user/code/debug/.venv/lib/python3.10/site-packages/vllm/v1/engine/async_llm.py", line 477, in output_handler
(APIServer pid=207278) ERROR 11-30 15:14:29 [async_llm.py:525] outputs = await engine_core.get_output_async()
(APIServer pid=207278) ERROR 11-30 15:14:29 [async_llm.py:525] File "/home/user/code/debug/.venv/lib/python3.10/site-packages/vllm/v1/engine/core_client.py", line 883, in get_output_async
(APIServer pid=207278) ERROR 11-30 15:14:29 [async_llm.py:525] raise self._format_exception(outputs) from None
(APIServer pid=207278) ERROR 11-30 15:14:29 [async_llm.py:525] vllm.v1.engine.exceptions.EngineDeadError: EngineCore encountered an issue. See stack trace (above) for the root cause.
(EngineCore_DP0 pid=207394) Traceback (most recent call last):
(EngineCore_DP0 pid=207394) File "/usr/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap
(EngineCore_DP0 pid=207394) self.run()
(EngineCore_DP0 pid=207394) File "/usr/lib/python3.10/multiprocessing/process.py", line 108, in run
(EngineCore_DP0 pid=207394) self._target(*self._args, **self._kwargs)
(EngineCore_DP0 pid=207394) File "/home/user/code/debug/.venv/lib/python3.10/site-packages/vllm/v1/engine/core.py", line 846, in run_engine_core
(EngineCore_DP0 pid=207394) raise e
(EngineCore_DP0 pid=207394) File "/home/user/code/debug/.venv/lib/python3.10/site-packages/vllm/v1/engine/core.py", line 835, in run_engine_core
(EngineCore_DP0 pid=207394) engine_core.run_busy_loop()
(EngineCore_DP0 pid=207394) File "/home/user/code/debug/.venv/lib/python3.10/site-packages/vllm/v1/engine/core.py", line 860, in | https://github.com/vllm-project/vllm/issues/29747 | closed | [
"bug"
] | 2025-11-30T13:20:23Z | 2025-12-02T22:42:30Z | 3 | hibukipanim |
vllm-project/vllm | 29,735 | [Usage]:Accessing free_blocks count from LLMEngine or LLM ? | ### Your current environment
```text
None
```
### How would you like to use vllm
I'm doing research on key-value caching optimization. I want to know how to determine the number of free blocks during runtime. I tried manually creating the engine, but I couldn't find the method after searching through the code.
AI keeps providing methods that have already been abandoned.
I would be very grateful for any help, as this has been puzzling me for hours.
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29735 | closed | [
"usage"
] | 2025-11-29T19:21:50Z | 2025-12-05T14:01:42Z | 4 | H-T-H |
vllm-project/vllm | 29,722 | [RFC]: Add Balance Scheduling | ### Motivation.
**Limitations of the current vLLM v1 scheduling strategy**
vLLM v1 scheduling currently enables chunkedprefill by default, which processes prefill and decode requests simultaneously in a single scheduling session. This can impact the overall system throughput and performance in some scenarios.
Balance scheduling addresses this issue by synchronizing the number of running queues across all schedulers to delay the scheduling of new requests, thereby improving the overall system's steady-state decoding time. This achieves:
✅Adding `balance_gather` to the scheduler synchronizes the number of requests in the running queues between DPs.
✅Balance scheduling improves the decode steady-state time, thereby increasing the overall output throughput of the inference system.
### Proposed Change.
**1.Feature Overview**
In the vLLM scheduler, running requests (i.e., requests that are already undergoing pre-filled computation) have the highest priority, followed by waiting requests (i.e., requests that have not yet been computed).
As shown in the diagram above, when the entire inference system exits from a steady state, the scheduler will schedule a batch of new requests for prefill operations and then synchronize them among the dynamic programming (DP) models. This can cause some DP models that are entirely decoded to synchronize with the number of prefilled tokens. Frequent prefill scheduling by certain DP models can lead to a deterioration in the overall system output throughput.
Balance scheduling synchronizes the number of running queue requests across different DPs, and only schedules new requests for prefilling when at least every scheduler has fewer than max_nun_requst.
**2.Implementation Design**
**3.Experiment Results**
- Fixed-length input scenario: In the performance test scenario with 3.5K fixed-length input and 1.5K fixed-length output, the throughput performance was improved by approximately **18%** after adding balance scheduling.
| Method | Model | Input Len | Request Count | Output Len | BatchSize | Average TTFT | Average TPOT | e2e duration | Input Token Throughput | Output Token Throughput | Request Throughput
| ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- |
| Baseline | DeepSeekV3.1 | 3500 | 512 | 1500 | 128 | 6600 | 86.85 | 591.9s | 3030.5 | 1297.3 | 0.86 |
| Balance scheduling | DeepSeekV3.1 | 3500 | 512 | 1500 | 128 | 7012 | 70.63 | 501.7s | 3575.7 | 1530.7 | 1.02 |
**4.Demo PR**
[#29721 ](https://github.com/vllm-project/vllm/pull/29721)
### Feedback Period.
No response
### CC List.
No response
### Any Other Things.
No response
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29722 | open | [
"RFC"
] | 2025-11-29T09:28:43Z | 2025-12-02T08:23:33Z | 0 | GDzhu01 |
vllm-project/vllm | 29,707 | [Usage]: Workaround to run model on GPUs with Compute Capability < 8.0? | ### Your current environment
Problem:
I am unable to run the Qwen3-VL-32B-Instruct-AWQ-4bit model due to a CUDA compute capability requirement. My hardware consists of two NVIDIA QUADRO RTX 5000 cards (16GB each, 32GB total) with a compute capability of 7.5. The software framework (likely a recent version of PyTorch or a specific library) raises an error:
"GPUs with compute capability < 8.0 are not supported."
Question:
Are there any workarounds to run this model on my older QUADRO RTX 5000 GPUs? Thanks in advance.
```
vllm collect-env
INFO 11-29 20:49:15 [__init__.py:216] Automatically detected platform cuda.
Collecting environment information...
==============================
System Info
==============================
OS : Ubuntu 24.04.3 LTS (x86_64)
GCC version : (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0
Clang version : Could not collect
CMake version : version 3.30.3
Libc version : glibc-2.39
==============================
PyTorch Info
==============================
PyTorch version : 2.8.0+cu128
Is debug build : False
CUDA used to build PyTorch : 12.8
ROCM used to build PyTorch : N/A
==============================
Python Environment
==============================
Python version : 3.12.11 | packaged by Anaconda, Inc. | (main, Jun 5 2025, 13:09:17) [GCC 11.2.0] (64-bit runtime)
Python platform : Linux-6.14.0-27-generic-x86_64-with-glibc2.39
==============================
CUDA / GPU Info
==============================
Is CUDA available : True
CUDA runtime version : 12.0.140
CUDA_MODULE_LOADING set to : LAZY
GPU models and configuration :
GPU 0: Quadro RTX 5000
GPU 1: Quadro RTX 5000
Nvidia driver version : 580.65.06
cuDNN version : Could not collect
HIP runtime version : N/A
MIOpen runtime version : N/A
Is XNNPACK available : True
==============================
CPU Info
==============================
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 46 bits physical, 48 bits virtual
Byte Order: Little Endian
CPU(s): 20
On-line CPU(s) list: 0-19
Vendor ID: GenuineIntel
Model name: Intel(R) Core(TM) i9-10900X CPU @ 3.70GHz
CPU family: 6
Model: 85
Thread(s) per core: 2
Core(s) per socket: 10
Socket(s): 1
Stepping: 7
CPU(s) scaling MHz: 28%
CPU max MHz: 4700.0000
CPU min MHz: 1200.0000
BogoMIPS: 7399.70
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cdp_l3 ssbd mba ibrs ibpb stibp ibrs_enhanced fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm mpx rdt_a avx512f avx512dq rdseed adx smap clflushopt clwb intel_pt avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req avx512_vnni md_clear flush_l1d arch_capabilities
L1d cache: 320 KiB (10 instances)
L1i cache: 320 KiB (10 instances)
L2 cache: 10 MiB (10 instances)
L3 cache: 19.3 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-19
Vulnerability Gather data sampling: Vulnerable
Vulnerability Ghostwrite: Not affected
Vulnerability Itlb multihit: KVM: Mitigation: VMX unsupported
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Mitigation; Enhanced IBRS
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Enhanced / Aut | https://github.com/vllm-project/vllm/issues/29707 | closed | [
"usage"
] | 2025-11-29T00:47:39Z | 2025-11-30T06:04:29Z | 5 | seasoncool |
vllm-project/vllm | 29,679 | [Usage]: Get request total time | ### Your current environment
```text
==============================
System Info
==============================
OS : Ubuntu 22.04.5 LTS (x86_64)
GCC version : (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version : Could not collect
CMake version : version 3.28.0
Libc version : glibc-2.35
==============================
PyTorch Info
==============================
PyTorch version : 2.9.0+cu128
Is debug build : False
CUDA used to build PyTorch : 12.8
ROCM used to build PyTorch : N/A
==============================
Python Environment
==============================
Python version : 3.10.19 | packaged by conda-forge | (main, Oct 22 2025, 22:29:10) [GCC 14.3.0] (64-bit runtime)
Python platform : Linux-6.8.0-1030-azure-x86_64-with-glibc2.35
==============================
CUDA / GPU Info
==============================
Is CUDA available : True
CUDA runtime version : Could not collect
CUDA_MODULE_LOADING set to :
GPU models and configuration : GPU 0: NVIDIA H100 NVL
Nvidia driver version : 535.247.01
cuDNN version : Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.9.10.2
/usr/lib/x86_64-linux-gnu/libcudnn_adv.so.9.10.2
/usr/lib/x86_64-linux-gnu/libcudnn_cnn.so.9.10.2
/usr/lib/x86_64-linux-gnu/libcudnn_engines_precompiled.so.9.10.2
/usr/lib/x86_64-linux-gnu/libcudnn_engines_runtime_compiled.so.9.10.2
/usr/lib/x86_64-linux-gnu/libcudnn_graph.so.9.10.2
/usr/lib/x86_64-linux-gnu/libcudnn_heuristic.so.9.10.2
/usr/lib/x86_64-linux-gnu/libcudnn_ops.so.9.10.2
HIP runtime version : N/A
MIOpen runtime version : N/A
Is XNNPACK available : True
==============================
CPU Info
==============================
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 48 bits physical, 48 bits virtual
Byte Order: Little Endian
CPU(s): 40
On-line CPU(s) list: 0-39
Vendor ID: AuthenticAMD
Model name: AMD EPYC 9V84 96-Core Processor
CPU family: 25
Model: 17
Thread(s) per core: 1
Core(s) per socket: 40
Socket(s): 1
Stepping: 1
BogoMIPS: 4800.05
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl tsc_reliable nonstop_tsc cpuid extd_apicid aperfmperf tsc_known_freq pni pclmulqdq ssse3 fma cx16 pcid sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw topoext perfctr_core vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves user_shstk avx512_bf16 clzero xsaveerptr rdpru arat avx512vbmi umip avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq rdpid fsrm
Hypervisor vendor: Microsoft
Virtualization type: full
L1d cache: 1.3 MiB (40 instances)
L1i cache: 1.3 MiB (40 instances)
L2 cache: 40 MiB (40 instances)
L3 cache: 160 MiB (5 instances)
NUMA node(s): 1
NUMA node0 CPU(s): 0-39
Vulnerability Gather data sampling: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Vulnerable: Safe RET, no microcode
Vulnerability Spec store bypass: Vulnerable
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Retpolines; STIBP disabled; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected
==============================
Versions of relevant libraries
==============================
[pip3] flashinfer-python==0.5.2
[pip3] numpy==1.26.4
[pip3] nvidia-cublas-cu12==12.8.4.1
[pip3] nvidia-cuda-cupti-cu12==12.8.90
[pip3] nvidia-cuda-nvrtc-cu12==12.8.93
[pip3] nvidia-cuda-runtime-cu12==12.8.90
| https://github.com/vllm-project/vllm/issues/29679 | closed | [
"usage"
] | 2025-11-28T14:03:16Z | 2025-12-01T09:34:12Z | 5 | chwundermsft |
huggingface/lerobot | 2,543 | Different finetune loss given policy.type=pi0 / policy.path=lerobot/pi0_base. What is the difference? | Hi, I have two different configurations:
1. ` --dataset.repo_id=BBBBBBob/libero_goal_lerobot \
--dataset.root=/home/j84403411/data/libero/libero_goal_lerobot \
--policy.path=lerobot/pi0_base \
--policy.push_to_hub=false \
--policy.use_proprio=true \
--output_dir=/home/j84403411/checkpoint/libero/pi0/libero_goal_proprio \
--policy.dtype=bfloat16 \
--steps=40_000 \
--batch_size=16 \
--rename_map='{"observation.images.image":"observation.images.base_0_rgb", "observation.images.wrist_image":"observation.images.left_wrist_0_rgb"}' \ `
and
2.
` --dataset.repo_id=BBBBBBob/libero_goal_lerobot \
--dataset.root=/home/j84403411/data/libero/libero_goal_lerobot \
--policy.type=pi0 \
--policy.pretrained_path=lerobot/pi0_base \
--policy.push_to_hub=false \
--policy.use_proprio=true \
--output_dir=/home/j84403411/checkpoint/libero/pi0/libero_goal_proprio \
--policy.dtype=bfloat16 \
--steps=40_000 \
--batch_size=16 \
--policy.input_features='{"observation.state": {"type": "STATE", "shape": [8]},
"observation.images.wrist_image": {"type": "VISUAL", "shape": [3, 256, 256]},
"observation.images.image": {"type": "VISUAL", "shape": [3, 256, 256]},
}' \
--policy.output_features='{"action": {"type": "ACTION", "shape": [7]}}' \ `
The loss trained from the second configuration is 10 times higher than the first one. What caused the difference? Do you know if different checkpoints are loaded in this case? I appreciate your help! | https://github.com/huggingface/lerobot/issues/2543 | closed | [] | 2025-11-28T12:34:38Z | 2025-12-01T11:25:17Z | null | BBBBBBob |
huggingface/transformers.js | 1,467 | Missing the following inputs: input_points, input_labels (or input_boxes) | ### Question
thanks for your excellent works!
I just write test code for SlimSAM model powered by transformers.js referring to this example(with some improvements): https://github.com/huggingface/transformers.js-examples/blob/main/segment-anything-webgpu/index.js
my code for `decode` method:
```js
// Decode segmentation
async function decode() {
if (!imageEmbeddings || isDecoding || isEncoding) return;
if (isDecoding) {
decodePending = true;
return;
}
isDecoding = true;
try {
let input_points = null;
let input_labels = null;
let input_boxes = null;
let outputs = null;
if (promptMode == "point" && points.length > 0) {
const reshaped = imageprocessed.reshaped_input_sizes[0]; // [H, W]
const scaledPoints = points.map(p => [
p.x * reshaped[1],
p.y * reshaped[0]
]);
const labels = points.map(p => BigInt(p.label));
input_points = new Tensor("float32", scaledPoints.flat(), [1, 1, points.length, 2]);
input_labels = new Tensor("int64", labels, [1, 1, points.length]);
// Fallback: if no prompts, skip
if (!input_points) return;
// Run model with point mode
outputs = await model({
...imageEmbeddings,
input_points: input_points,
input_labels: input_labels,
input_boxes: null
});
}
if (promptMode == "box" && box) {
const reshaped = imageprocessed.reshaped_input_sizes[0];
const [x1, y1, x2, y2] = [
box.x1 * reshaped[1],
box.y1 * reshaped[0],
box.x2 * reshaped[1],
box.y2 * reshaped[0]
];
input_boxes = new Tensor("float32", [x1, y1, x2, y2], [1, 1, 4]);
// Fallback: if no prompts, skip
if (!input_boxes) return;
// Run model with box mode
outputs = await model({
...imageEmbeddings,
input_points: null,
input_labels: null,
input_boxes: input_boxes
});
}
// Post-process
const masks = await processor.post_process_masks(
outputs.pred_masks,
imageprocessed.original_sizes,
imageprocessed.reshaped_input_sizes
);
const scores = outputs.iou_scores.data;
updateMask(masks[0], scores); // masks[0] is [3, H, W]
} catch (e) {
console.error("Decode error:", e);
statusEl.textContent = "❌ Segmentation failed.";
} finally {
isDecoding = false;
if (decodePending) {
decodePending = false;
decode();
}
}
}
```
it supports 2 prompt modes: `point` &` box` which selected by users on UI elements (html not provided).
but error printed every time when running `decode` method (at the line of calling `outputs = await model(...)`), the error message is:
with box prompt mode:
`Error: An error occurred during model execution: "Missing the following inputs: input_points, input_labels.`
with point prompt mode:
`Error: An error occurred during model execution: "Missing the following inputs: input_boxes.`
Should I pass all three parameters(input_points/input_labels/input_boxes) simultaneously, regardless of which prompt mode I’m using? How could I support point & box at the same time, since no demo codes found on internet. thanks!
```
version: transformers.js 3.5.0 from https://cdn.jsdelivr.net/npm/@huggingface/transformers@3.5.0
os: Windows 10
chorme: 142
model: Xenova/slimsam-77-uniform
``` | https://github.com/huggingface/transformers.js/issues/1467 | closed | [
"question"
] | 2025-11-28T10:01:04Z | 2025-12-01T04:04:59Z | null | sherlockchou86 |
vllm-project/vllm | 29,643 | [Usage]: Enabling Tool call in the Python SDK | ### Your current environment
Hi Team,
I am currently exploring VLLM to enable tool calling, and I need some support with this. It would be very helpful if you could provide the corresponding Python code.
What I’m trying to achieve is to configure the Python package with the same settings that I use when starting the VLLM server. The configuration I’m using is:
vllm serve DeepSeek-R1-0528-Qwen3-8B \
--served-model-name deepseek \
--gpu_memory_utilization 0.5 \
--max_num_seqs 20 \
--max_model_len 10000 \
--enable-auto-tool-choice \
--tool-call-parser deepseek_v3 \
--chat-template tool_chat_template_deepseekr1.jinja \
--port 5050 \
--max_num_batched_tokens 5000
I need to replicate this exact configuration in Python.
Your support would be greatly appreciated. Please respond at your earliest convenience.
If you want, I can also write the **Python code equivalent** for these VLLM configurations.
Best Regards
Madan
### How would you like to use vllm
I want to use vLLM to serve a model with tool-calling support enabled. Specifically, I need to run the model with the same configuration parameters that I currently use when launching the vLLM server from the command line. These settings include GPU memory utilization, maximum sequence limits, tool-calling options, a custom tool-call parser, and a custom chat template.
My goal is to reproduce the following server configuration within a Python environment using the vLLM Python API:
vllm serve DeepSeek-R1-0528-Qwen3-8B \
--served-model-name deepseek \
--gpu_memory_utilization 0.5 \
--max_num_seqs 20 \
--max_model_len 10000 \
--enable-auto-tool-choice \
--tool-call-parser deepseek_v3 \
--chat-template tool_chat_template_deepseekr1.jinja \
--port 5050 \
--max_num_batched_tokens 5000
`
In short, I need Python code that sets these exact configurations so I can run vLLM programmatically with tool calling enabled.
If you want, I can also provide the **Python code equivalent** for this configuration.
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29643 | open | [
"usage"
] | 2025-11-28T04:39:47Z | 2025-12-01T14:54:47Z | 2 | Madan1215 |
vllm-project/vllm | 29,641 | [Bug]: Max Tokens not being honoured in Chat Completions for GPTOSS model | ### Your current environment
It seems that in the latest version of vllm 0.11+ Chat Completions has stopped honouring `max_tokens` with GPTOSS 120B model, the below request payload has stopped working with `max_tokens` earlier the same payload would provide an output to the limit of the `max_tokens` provided..
Interestingly if you look at the `usage` tokens, it's showing `completion_tokens` as 500 but the output is BLANK.
```json
{
"messages": [
{
"role": "user",
"content": "What is the role of AI in medicine?"
}
],
"model": "openai/gpt-oss-120b",
"max_tokens": 500,
"reasoning": {"effort": "low"},
"stream": false
}
```
getting BLANK output, even though the `usage` is showing token counts created is matching max_tokens
```json
{
"id": "chatcmpl-c71e934ac0b74bd4b8f99fe9b5516ea3",
"object": "chat.completion",
"created": 1764300020,
"model": "openai/gpt-oss-120b",
"choices": [
{
"index": 0,
"message": {
"role": "assistant",
"content": null,
"refusal": null,
"annotations": null,
"audio": null,
"function_call": null,
"tool_calls": [],
"reasoning": "Need to answer.",
"reasoning_content": "Need to answer."
},
"logprobs": null,
"finish_reason": "length",
"stop_reason": null,
"token_ids": null
}
],
"service_tier": null,
"system_fingerprint": null,
"usage": {
"prompt_tokens": 78,
"total_tokens": 578,
"completion_tokens": 500,
"prompt_tokens_details": null
},
"prompt_logprobs": null,
"prompt_token_ids": null,
"kv_transfer_params": null
}
```
When you remove the `max_tokens`, we get the output which shows `usage_token` to have `completion_tokens` to be around 1600 tokens..
It seems that starting from vllm 0.11+ version, the auto-truncation using the `max_tokens` has stopped working
```json
{
"id": "chatcmpl-61b60144d43147e2b007158712ad4920",
"object": "chat.completion",
"created": 1764300423,
"model": "openai/gpt-oss-120b",
"choices": [
{
"index": 0,
"message": {
"role": "assistant",
"content": "**The role of AI in medicine is expanding rapidly and touches virtually every aspect of healthcare—from the way doctors diagnose patients to how hospitals run their operations.** Below is a structured overview that covers the major domains, concrete examples, benefits, challenges, and future directions.\n\n---\n\n## 1. Clinical Care\n\n| Sub‑area | What AI Does | Real‑World Examples | Benefits |\n|----------|--------------|---------------------|----------|\n| **Diagnostics** | Image analysis, pattern recognition, risk stratification | • Radiology: Google DeepMind’s AI detects lung cancer on CT scans with >95% accuracy.<br>• Dermatology: FDA‑cleared apps (e.g., SkinVision) classify skin lesions from photos.<br>• Pathology: Paige.ai assists in detecting prostate cancer in biopsy slides. | Faster, more consistent readings; can catch subtle findings that human eyes miss. |\n| **Predictive Analytics** | Forecast disease onset, complications, readmission risk | • Sepsis prediction models (e.g., Epic Sepsis Model) trigger alerts hours before clinical signs.<br>• Cardiovascular risk calculators incorporating genomics and wearables. | Enables proactive interventions, reduces morbidity and cost. |\n| **Treatment Planning** | Decision support, dose optimisation, drug selection | • IBM Watson for Oncology (clinical trial matching).<br>• Radiation oncology: AI‑driven dose‑painting to spare healthy tissue.<br>• Pharmacogenomics: AI predicts drug‑gene interactions. | Personalises therapy, improves outcomes, reduces adverse events. |\n| **Robotics & Minimally Invasive Surgery** | Real‑time image guidance, autonomous suturing, task automation | • Da Vinci Surgical System (augmented with AI for instrument tracking).<br>• VERDICT AI for autonomous suturing in animal models. | Increases precision, reduces surgeon fatigue, shortens recovery. |\n\n---\n\n## 2. Patient‑Facing Applications\n\n| Application | Description | Example |\n|-------------|-------------|---------|\n| **Virtual Assistants & Chatbots** | Symptom triage, medication reminders, mental‑health chat | • Babylon Health (AI‑driven triage).<br>• Woebot (CBT‑based mental‑health chatbot). |\n| **Telemedicine Enhancements** | Real‑time vitals extraction from video, automated note‑taking | • KardiaMobile ECG integration with AI‑based arrhythmia detection. |\n| **Wearables & Remote Monitoring** | Continuous data streams analysed for early alerts | • Apple Watch ECG + AI arrhythmia detection; Fitbit heart‑rate trend alerts. |\n\n---\n\n## 3. Operational & Administrative Efficiency\n\n| Domain | AI Functions | Example | | https://github.com/vllm-project/vllm/issues/29641 | closed | [
"bug"
] | 2025-11-28T03:39:34Z | 2025-12-21T02:39:32Z | 16 | soodrohit |
huggingface/transformers | 42,464 | Add SAM 3D Objects Encoder | ### Model description
## Model Description
SAM 3D Objects is Meta AI's foundation model for 3D object reconstruction from single images. I'm proposing to add the **encoder component** (DINOv2-based Vision Transformer) to Transformers.
**Scope**: Encoder only, not the full 3D generation pipeline (which includes Gaussian Splatting/Mesh decoders better suited for Diffusers).
## Open source status
- [x] The model implementation available
- [x] The model weights are available
## Provide useful links for the implementation
- **Model Card**: https://huggingface.co/facebook/sam-3d-objects
- **Paper**: https://arxiv.org/abs/2511.16624
- **Original Repository**: https://github.com/facebookresearch/sam-3d-objects
- **Blog Post**: https://ai.meta.com/blog/sam-3d/
## Implementation Progress
I have already implemented this model and it's ready for review:
✅ **Implementation Complete:**
- `Sam3DObjectsEncoderConfig` - Configuration with DINO variant support
- `Sam3DObjectsEncoder` - Main encoder model
- `Sam3DObjectsEncoderForMasks` - Variant for mask encoding
- `Sam3DObjectsImageProcessor` - Image preprocessing
- Comprehensive test suite: **28/28 tests passing**
- Full documentation
**Test Results:**
collected 29 items
28 passed, 1 skipped in 4.92s
**Example Usage:**
```python
from transformers.models.sam3d_objects import (
Sam3DObjectsEncoder,
Sam3DObjectsEncoderConfig,
Sam3DObjectsImageProcessor,
)
config = Sam3DObjectsEncoderConfig.from_dino_config("dinov2_vitl14")
model = Sam3DObjectsEncoder(config)
processor = Sam3DObjectsImageProcessor()
inputs = processor(images=image, return_tensors="pt")
outputs = model(**inputs)
embeddings = outputs.last_hidden_state
```
## Questions
1. Is there interest in adding the SAM 3D Objects Encoder to Transformers?
2. Should this be limited to the encoder component (my recommendation)?
3. Should I submit a PR, or are there any requirements I should address first?
## Additional Context
- The encoder is based on DINOv2 and fits naturally in Transformers
- Full 3D generation pipeline would be better suited for Diffusers
- Model is gated on Hub (requires license acceptance)
- Implementation follows Transformers patterns and guidelines
I'm ready to submit a PR and address any feedback.
### Open source status
- [x] The model implementation is available
- [x] The model weights are available
### Provide useful links for the implementation
## Links
- **Model Card**: https://huggingface.co/facebook/sam-3d-objects
- **Paper**: https://arxiv.org/abs/2511.16624 (SAM 3D: 3Dfy Anything in Images)
- **Original Repository**: https://github.com/facebookresearch/sam-3d-objects
- **Blog Post**: https://ai.meta.com/blog/sam-3d/
- **Project Page**: https://ai.meta.com/sam3d/
## Authors
**SAM 3D Team** from Meta AI
For the complete author list and contributions, see:
- [ArXiv Paper](https://arxiv.org/abs/2511.16624)
- [Original Repository](https://github.com/facebookresearch/sam-3d-objects)
*Note: This is a large collaborative project with many contributors from Meta Superintelligence Labs.*
## Implementation Details
**Model Type**: Vision Encoder (DINOv2-based)
**Architecture**: Vision Transformer (ViT)
**Variants Supported**:
- ViT-S/14 (384 dim)
- ViT-B/14 (768 dim)
- ViT-L/14 (1024 dim)
- ViT-G/14 (1536 dim)
**Input**: RGB images (224x224 or 518x518)
**Output**: Visual embeddings for 3D generation tasks
**License**: SAM License (gated model on HuggingFace Hub) | https://github.com/huggingface/transformers/issues/42464 | open | [
"New model"
] | 2025-11-27T19:48:28Z | 2025-12-05T10:32:33Z | 1 | Aznix07 |
vllm-project/vllm | 29,584 | [Usage]: Can KV Cache be disabled in non-autoregressive generation tasks? | ### Your current environment
```text
==============================
System Info
==============================
OS : Ubuntu 24.04.3 LTS (x86_64)
GCC version : (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0
Clang version : Could not collect
CMake version : version 3.28.3
Libc version : glibc-2.39
==============================
PyTorch Info
==============================
PyTorch version : 2.9.0+cu128
Is debug build : False
CUDA used to build PyTorch : 12.8
ROCM used to build PyTorch : N/A
==============================
Python Environment
==============================
Python version : 3.10.15 (main, Oct 3 2024, 07:27:34) [GCC 11.2.0] (64-bit runtime)
Python platform : Linux-6.8.0-87-generic-x86_64-with-glibc2.39
==============================
CUDA / GPU Info
==============================
Is CUDA available : True
CUDA runtime version : Could not collect
CUDA_MODULE_LOADING set to :
GPU models and configuration :
GPU 0: NVIDIA GeForce RTX 3090
GPU 1: NVIDIA GeForce RTX 3090
Nvidia driver version : 575.57.08
cuDNN version : Could not collect
HIP runtime version : N/A
MIOpen runtime version : N/A
Is XNNPACK available : True
==============================
vLLM Info
==============================
ROCM Version : Could not collect
vLLM Version : 0.11.2
vLLM Build Flags:
CUDA Archs: Not Set; ROCm: Disabled
GPU Topology:
GPU0 GPU1 CPU Affinity NUMA Affinity GPU NUMA ID
GPU0 X SYS 0-23,48-71 0 N/A
GPU1 SYS X 24-47,72-95 1 N/A
Legend:
X = Self
SYS = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
PHB = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
PXB = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
PIX = Connection traversing at most a single PCIe bridge
NV# = Connection traversing a bonded set of # NVLinks
```
### How would you like to use vllm
Hello vLLM team,
Currently, vLLM (v0.11.2) enables KV cache for certain LLM-based pooling and reranking models, such as the Qwen3-Embedding series, even when `--no-enable-chunked-prefill` and `--no-enable-prefix-caching` are set. This leads to unnecessary GPU memory usage.
Would it be possible to disable KV cache for pooling and reranking models under these conditions?
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29584 | open | [
"usage"
] | 2025-11-27T05:30:08Z | 2025-12-05T02:40:28Z | 5 | GitEventhandler |
vllm-project/vllm | 29,574 | [Performance]: Using vLLM to accelerate VLM models, does the vision encoding part currently support parallel processing, or is it still being processed serially? | ### Proposal to improve performance
I found that currently, images of different sizes are processed sequentially, which significantly slows down the processing speed. How can we adapt to parallel processing? Should we resize or pad all images to the same size for batch processing, or can we run multiple encoder models in parallel? Thank you.
### Report of performance regression
_No response_
### Misc discussion on performance
_No response_
### Your current environment (if you think it is necessary)
```text
The output of `python collect_env.py`
```
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29574 | open | [
"performance"
] | 2025-11-27T03:51:36Z | 2025-11-27T10:54:09Z | 2 | NewZxy |
vllm-project/vllm | 29,564 | [Doc]: Make PyTorch profiler gzip and CUDA time dump configurable | ### 📚 The doc issue
We observed that enabling both use_gzip and dump_self_cuda_time_total in the vLLM torch profiler introduces significant overhead during profiling.
For example, when profiling 10 randomly generated requests (1000 input tokens, 200 output tokens) on an A100 using the Qwen3-32B model, we found that gzip compression of the profiling trace and dumping the CUDA time table take ~68 seconds, dominating the overall profiling time.
The main sources of overhead appear to be:
1. Gzip compression of the profiling trace file
2. Generation and dumping of the CUDA time summary table
After disabling these two features, the total profiling dump time is reduced to ~18 seconds.
In many profiling scenarios (e.g., quick performance checks or small-scale experiments), users may not need gzip compression or the CUDA time table. Therefore, it would be helpful to make these two behaviors individually configurable via environment variables—enabled by default for completeness, but optionally turnable off when faster profiling turnaround is preferred. Moreover, gzip compression could potentially be performed asynchronously after the trace is dumped, allowing lower-latency profiling in staging or pre-production environments.
This patch proposes adding such configurability so users can selectively disable gzip compression and/or CUDA time table generation when they want a faster and lighter profiling workflow.
### Suggest a potential alternative/fix
_No response_
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29564 | closed | [
"documentation"
] | 2025-11-27T02:21:20Z | 2025-12-01T04:30:48Z | 1 | zhangruoxu |
vllm-project/vllm | 29,562 | [Bug]: "\n\n" content between reasoning and tool_call content when tool_call and stream mode | ### Your current environment
<details>
<summary>The output of <code>python collect_env.py</code></summary>
```text
Your output of `python collect_env.py` here
```
</details>
### 🐛 Describe the bug
https://github.com/QwenLM/Qwen3/issues/1755
When stream mode true, the response contains content "\n\n" between reasoning and tool_call; but with stream model false, it didn't generate content "\n\n".
Is there some thing different, I don't want the content "\n\n" between reasoning and tool_call.
<img width="974" height="533" alt="Image" src="https://github.com/user-attachments/assets/0cc36343-3c0f-4ce1-9028-30f561a55dac" />
Here is my requests:
```
{
"model": "Qwen3-235B-A22B-Thinking-2507",
"tools": [
{
"type": "function",
"function": {
"name": "search_law_articles",
"parameters": {
"type": "object",
"properties": {
"level": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "搜索条件：法规类型"
},
"query": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "查询语句"
},
"title": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "法律标题"
},
"article": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "法律条款序号，如 第十条"
},
"content": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "法律条款及内容，如 第十条 贷款人委托支付"
},
"pub_department": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "发布部门"
},
"pub_time_after": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "搜索条件：发布时间晚于此时间，格式如2025-06-20"
},
"pub_time_before": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "搜索条件：发布时间早于此时间，格式如2025-06-20"
},
"imply_time_after": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "搜索条件：实施时间晚于此时间，格式如2025-06-20"
},
"imply_time_before": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "搜索条件：实施时间早于此时间，格式如2025-06-20"
}
}
},
"description": "此工具用于搜索法条内容， 库中是按照法律条目进行存储， 查询可选多个查询过滤条件"
}
}
],
"stream": true,
"messages": [
{
"role": "user",
"content": [
{
"text": "帮我解读下网络安全法",
"type": "text"
}
]
}
]
}
```
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29562 | open | [
"bug"
] | 2025-11-27T01:49:04Z | 2025-11-27T01:49:04Z | 0 | NiuBlibing |
vllm-project/vllm | 29,560 | [Doc]: Batch Invariance on Ampere Platforms | ### 📚 The doc issue
Does the batch invariance feature released in vllm 0.11.2 support the Ampere architecture? If adaptations are required, what modifications need to be made?
### Suggest a potential alternative/fix
_No response_
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29560 | closed | [
"documentation"
] | 2025-11-27T01:06:49Z | 2025-11-27T14:21:30Z | 0 | luo1206 |
huggingface/trl | 4,582 | Does the GRPO Trainer support multi-image input for Qwen3-VL? | Does the GRPO Trainer support multi-image input for Qwen3-VL? | https://github.com/huggingface/trl/issues/4582 | open | [
"🏋 GRPO"
] | 2025-11-26T14:03:57Z | 2025-11-27T08:08:25Z | 1 | Lestoky |
huggingface/diffusers | 12,722 | How to run qwen-image in kaggle gpu T4 * 2 successfully? | ```python3
!python3 -m pip install -U diffusers peft bitsandbytes
import diffusers, torch, math
qwen = diffusers.QwenImagePipeline.from_pretrained('Qwen/Qwen-Image', torch_dtype=torch.float16, low_cpu_mem_usage=True, quantization_config=diffusers.PipelineQuantizationConfig(quant_backend='bitsandbytes_4bit', quant_kwargs={'load_in_4bit':True, 'bnb_4bit_quant_type':'nf4', 'bnb_4bit_compute_dtype':torch.float16}, components_to_quantize=['transformer', 'text_encoder']))
qwen.scheduler = diffusers.FlowMatchEulerDiscreteScheduler.from_config({'base_image_seq_len':256, 'base_shift':math.log(3), 'invert_sigmas':False, 'max_image_seq_len':8192, 'max_shift':math.log(3), 'num_train_timesteps':1000, 'shift':1, 'shift_terminal':None, 'stochastic_sampling':False, 'time_shift_type':'exponential', 'use_beta_sigmas':False, 'use_dynamic_shifting':True, 'use_exponential_sigmas':False, 'use_karras_sigmas':False})
qwen.load_lora_weights('lightx2v/Qwen-Image-Lightning', weight_name='Qwen-Image-Lightning-4steps-V2.0.safetensors', adapter_name='lightning')
qwen.set_adapters('lightning', adapter_weights=1)
qwen.enable_sequential_cpu_offload()
qwen(prompt='a beautiful girl', height=1280, width=720, num_inference_steps=4, true_cfg_scale=1).images[0].save('a.png')
```
----> 3 qwen = diffusers.QwenImagePipeline.from_pretrained('Qwen/Qwen-Image', torch_dtype=torch.float16, low_cpu_mem_usage=True, quantization_config=diffusers.PipelineQuantizationConfig(quant_backend='bitsandbytes_4bit', quant_kwargs={'load_in_4bit':True, 'bnb_4bit_quant_type':'nf4', 'bnb_4bit_compute_dtype':torch.float16}, components_to_quantize=['transformer', 'text_encoder']))
OutOfMemoryError: CUDA out of memory. Tried to allocate 34.00 MiB. GPU 0 has a total capacity of 14.74 GiB of which 4.19 MiB is free. Process 8568 has 14.73 GiB memory in use. Of the allocated memory 14.50 GiB is allocated by PyTorch, and 129.00 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)
How to get more cuda memory?
@yiyixuxu @DN6 | https://github.com/huggingface/diffusers/issues/12722 | open | [] | 2025-11-26T12:53:30Z | 2025-11-28T03:54:07Z | null | chaowenguo |
vllm-project/vllm | 29,494 | [Doc]: Documentation inconsistency: Blog mentions append_slots() but codebase uses allocate_slots() | ### 📚 The doc issue
The Automatic Prefix Caching blog post mentions:
> "The scheduler calls kv_cache_manager.append_slots()"
However, the actual codebase uses a unified `kv_cache_manager.allocate_slots()` method that handles both prefill and decode requests.
**Location:**
- Blog: [[link to blog post](https://docs.vllm.ai/en/v0.8.5/design/v1/prefix_caching.html#operations)]
- Code: ./vllm/v1/core/kv_cache_manager.py
### Suggest a potential alternative/fix
Update the blog post to reflect the actual implementation `kv_cache_manager.allocate_slots()`
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29494 | closed | [
"documentation"
] | 2025-11-26T11:37:40Z | 2025-11-26T11:46:08Z | 1 | pradsgit |
huggingface/transformers | 42,418 | Custom nn.Parameter initialization in PreTrainedModel subclasses is overwritten by post_init()/from_pretrained() causing NaNs/Zeros | ### System Info
- `transformers` version: 4.57.1
- Platform: Linux-4.18.0-147.mt20200626.413.el8_1.x86_64-x86_64-with-glibc2.35
- Python version: 3.10.14
- Huggingface_hub version: 0.35.3
- Safetensors version: 0.6.2
- Accelerate version: 1.11.0
- Accelerate config: not found
- DeepSpeed version: 0.18.2
- PyTorch version (accelerator?): 2.7.1+cu126 (CUDA)
- Tensorflow version (GPU?): not installed (NA)
- Flax version (CPU?/GPU?/TPU?): not installed (NA)
- Jax version: not installed
- JaxLib version: not installed
- Using distributed or parallel set-up in script?: No
- Using GPU in script?: No
- GPU type: NVIDIA A100-SXM4-80GB
### Who can help?
@Cyrilvallez @zucchini-nlp
### Information
- [ ] The official example scripts
- [x] My own modified scripts
### Tasks
- [ ] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [x] My own task or dataset (give details below)
### Reproduction
```python
import numpy as np
import os
import random
import torch
import torch.nn as nn
from transformers import Qwen3VLForConditionalGeneration
def seed_everything(TORCH_SEED):
random.seed(TORCH_SEED)
os.environ["PYTHONHASHSEED"] = str(TORCH_SEED)
np.random.seed(TORCH_SEED)
torch.manual_seed(TORCH_SEED)
torch.cuda.manual_seed(TORCH_SEED)
torch.cuda.manual_seed_all(TORCH_SEED)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
seed_everything(66)
class TestModel1(Qwen3VLForConditionalGeneration):
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.action_head = nn.Linear(1024, 7)
self.positional_embedding = nn.Parameter(torch.randn(16, 1152))
self.post_init()
class TestModel2(nn.Module):
def __init__(self, *args, model_path, **kwargs):
super().__init__(*args, **kwargs)
self.model = Qwen3VLForConditionalGeneration.from_pretrained(model_path)
self.action_head = nn.Linear(1024, 7)
self.positional_embedding = nn.Parameter(torch.randn(16, 1152))
test_model1 = TestModel1.from_pretrained("Qwen/Qwen3-VL-4B-Instruct")
test_model2 = TestModel2(model_path="Qwen/Qwen3-VL-4B-Instruct")
print(test_model1.positional_embedding)
print(test_model1.positional_embedding.mean(), test_model1.positional_embedding.std())
print(test_model2.positional_embedding)
print(test_model2.positional_embedding.mean(), test_model2.positional_embedding.std())
````
### Expected behavior
When subclassing a model (inheriting from PreTrainedModel, e.g., Qwen3VLForConditionalGeneration, LlamaForCausalLM) to add custom learnable parameters, user-defined initialization in __init__ is often silently overwritten.
This occurs because from_pretrained (or the end of __init__) triggers self.post_init(), which recursively calls _init_weights. This mechanism re-initializes all parameters, ignoring the explicit initialization code provided by the user in __init__.
In the specific case of Qwen3-VL (and potentially others), this re-initialization results in NaNs or Zeros, rendering the model unusable without manual intervention.
Steps to reproduce The following script demonstrates the issue. Note: I used torch.randn for the custom parameter initialization. While I understand that torch.randn samples from a standard normal distribution and does not guarantee an exact sample mean of 0 and std of 1, it should result in valid float values. The observed NaNs/Zeros confirm that this initialization is being discarded and replaced by a faulty internal initialization logic. | https://github.com/huggingface/transformers/issues/42418 | open | [
"Usage",
"Feature request",
"bug"
] | 2025-11-26T10:29:57Z | 2025-12-01T15:10:32Z | 10 | Noietch |
huggingface/diffusers | 12,720 | how to quantization wan 2.2 vace after loading lora? | ```python3
diffusers.WanVACEPipeline.from_pretrained('linoyts/Wan2.2-VACE-Fun-14B-diffusers', vae=diffusers.AutoencoderKLWan.from_pretrained('linoyts/Wan2.2-VACE-Fun-14B-diffusers', subfolder='vae', torch_dtype=torch.float32), torch_dtype=torch.bfloat16, quantization_config=diffusers.PipelineQuantizationConfig(quant_backend='bitsandbytes_8bit', quant_kwargs={'load_in_8bit':True}, components_to_quantize=['transformer', 'transformer_2'])).save_pretrained('wan')
```
normally I can save the quantization model in this way
But now I want to merge lora and the quantization and then save the model with lora. How?
```python3
wan = diffusers.WanVACEPipeline.from_pretrained('linoyts/Wan2.2-VACE-Fun-14B-diffusers', vae=diffusers.AutoencoderKLWan.from_pretrained('linoyts/Wan2.2-VACE-Fun-14B-diffusers', subfolder='vae', torch_dtype=torch.float32), torch_dtype=torch.bfloat16)
wan.load_lora_weights('lightx2v/Wan2.2-Lightning', weight_name='Wan2.2-I2V-A14B-4steps-lora-rank64-Seko-V1/high_noise_model.safetensors', adapter_name='lightning')
wan.load_lora_weights('lightx2v/Wan2.2-Lightning', weight_name='Wan2.2-I2V-A14B-4steps-lora-rank64-Seko-V1/low_noise_model.safetensors', adapter_name='lightning_2', load_into_transformer_2=True)
wan.set_adapters(['lightning', 'lightning_2'], adapter_weights=[1] * 2)
how to quantization and save_pretrained?
```
@yiyixuxu @DN6 | https://github.com/huggingface/diffusers/issues/12720 | open | [] | 2025-11-26T10:11:38Z | 2025-12-11T17:29:30Z | null | chaowenguo |
vllm-project/vllm | 29,489 | [Usage]: Removing last generated token from output and kv cache | ### Your current environment
```text
Collecting environment information...
==============================
System Info
==============================
OS : Ubuntu 24.04.3 LTS (x86_64)
GCC version : (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0
Clang version : Could not collect
CMake version : version 3.28.3
Libc version : glibc-2.39
==============================
PyTorch Info
==============================
PyTorch version : 2.9.0+cu128
Is debug build : False
CUDA used to build PyTorch : 12.8
ROCM used to build PyTorch : N/A
==============================
Python Environment
==============================
Python version : 3.13.5 | packaged by conda-forge | (main, Jun 16 2025, 08:27:50) [GCC 13.3.0] (64-bit runtime)
Python platform : Linux-6.8.0-87-generic-x86_64-with-glibc2.39
==============================
CUDA / GPU Info
==============================
Is CUDA available : True
CUDA runtime version : 12.8.93
CUDA_MODULE_LOADING set to :
GPU models and configuration :
GPU 0: NVIDIA B200
GPU 1: NVIDIA B200
GPU 2: NVIDIA B200
GPU 3: NVIDIA B200
GPU 4: NVIDIA B200
GPU 5: NVIDIA B200
GPU 6: NVIDIA B200
GPU 7: NVIDIA B200
Nvidia driver version : 570.195.03
cuDNN version : Could not collect
HIP runtime version : N/A
MIOpen runtime version : N/A
Is XNNPACK available : True
==============================
CPU Info
==============================
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 52 bits physical, 57 bits virtual
Byte Order: Little Endian
CPU(s): 224
On-line CPU(s) list: 0-223
Vendor ID: GenuineIntel
Model name: INTEL(R) XEON(R) PLATINUM 8570
CPU family: 6
Model: 207
Thread(s) per core: 2
Core(s) per socket: 56
Socket(s): 2
Stepping: 2
CPU(s) scaling MHz: 33%
CPU max MHz: 4000.0000
CPU min MHz: 800.0000
BogoMIPS: 4200.00
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 intel_ppin cdp_l2 ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect user_shstk avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req vnmi avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid bus_lock_detect cldemote movdiri movdir64b enqcmd fsrm md_clear serialize tsxldtrk pconfig arch_lbr ibt amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities ibpb_exit_to_user
Virtualization: VT-x
L1d cache: 5.3 MiB (112 instances)
L1i cache: 3.5 MiB (112 instances)
L2 cache: 224 MiB (112 instances)
L3 cache: 600 MiB (2 instances)
NUMA node(s): 2
NUMA node0 CPU(s): 0-55,112-167
NUMA node1 CPU(s): 56-111,168-223
Vulnerability Gather data sampling: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Enhanced / Automatic IBRS; IBPB conditional; RSB filling; PBRSB-eIBRS SW sequence; BHI | https://github.com/vllm-project/vllm/issues/29489 | open | [
"usage"
] | 2025-11-26T09:35:37Z | 2025-11-26T09:36:37Z | 0 | josefdra |
huggingface/diffusers | 12,719 | how to use quantization and device_map=balance to run qwen-image on kaggle T4 * 2 | ```python3
!python3 -m pip install -U diffusers peft bitsandbytes protobuf
import diffusers, torch, math
qwen = diffusers.QwenImagePipeline.from_pretrained('Qwen/Qwen-Image', quantization_config=diffusers.PipelineQuantizationConfig(quant_backend='bitsandbytes_4bit', quant_kwargs={'load_in_4bit':True, 'bnb_4bit_quant_type':'nf4', 'bnb_4bit_compute_dtype':torch.float16}, components_to_quantize=['transformer', 'text_encoder']), torch_dtype=torch.float16, device_map='balanced')
print(qwen.hf_device_map)
qwen.scheduler = diffusers.FlowMatchEulerDiscreteScheduler.from_config({'base_image_seq_len':256, 'base_shift':math.log(3), 'invert_sigmas':False, 'max_image_seq_len':8192, 'max_shift':math.log(3), 'num_train_timesteps':1000, 'shift':1, 'shift_terminal':None, 'stochastic_sampling':False, 'time_shift_type':'exponential', 'use_beta_sigmas':False, 'use_dynamic_shifting':True, 'use_exponential_sigmas':False, 'use_karras_sigmas':False})
qwen.load_lora_weights('lightx2v/Qwen-Image-Lightning', weight_name='Qwen-Image-Lightning-4steps-V2.0.safetensors', adapter_name='lightning')
qwen.set_adapters('lightning', adapter_weights=1)
qwen(prompt='a beautiful girl', height=1280, width=720, num_inference_steps=4, true_cfg_scale=1).images[0].save('a.png')
```
WARNING:accelerate.big_modeling:Some parameters are on the meta device because they were offloaded to the cpu.
{'text_encoder': 'cpu', 'vae': 0} where is the transformer ?
NotImplementedError: Cannot copy out of meta tensor; no data!
I want to ask how to make the above code work in kaggle. why 16G * 2 vram still not enough to run q4 quantization qwen-image? I want to take full advantage of 2 gpu. Do I need max_memory?
full error logs:
/usr/local/lib/python3.11/dist-packages/torch/utils/_contextlib.py in decorate_context(*args, **kwargs)
114 def decorate_context(*args, **kwargs):
115 with ctx_factory():
--> 116 return func(*args, **kwargs)
117
118 return decorate_context
/usr/local/lib/python3.11/dist-packages/diffusers/pipelines/qwenimage/pipeline_qwenimage.py in __call__(self, prompt, negative_prompt, true_cfg_scale, height, width, num_inference_steps, sigmas, guidance_scale, num_images_per_prompt, generator, latents, prompt_embeds, prompt_embeds_mask, negative_prompt_embeds, negative_prompt_embeds_mask, output_type, return_dict, attention_kwargs, callback_on_step_end, callback_on_step_end_tensor_inputs, max_sequence_length)
566 )
567 do_true_cfg = true_cfg_scale > 1 and has_neg_prompt
--> 568 prompt_embeds, prompt_embeds_mask = self.encode_prompt(
569 prompt=prompt,
570 prompt_embeds=prompt_embeds,
/usr/local/lib/python3.11/dist-packages/diffusers/pipelines/qwenimage/pipeline_qwenimage.py in encode_prompt(self, prompt, device, num_images_per_prompt, prompt_embeds, prompt_embeds_mask, max_sequence_length)
252
253 if prompt_embeds is None:
--> 254 prompt_embeds, prompt_embeds_mask = self._get_qwen_prompt_embeds(prompt, device)
255
256 prompt_embeds = prompt_embeds[:, :max_sequence_length]
/usr/local/lib/python3.11/dist-packages/diffusers/pipelines/qwenimage/pipeline_qwenimage.py in _get_qwen_prompt_embeds(self, prompt, device, dtype)
203 txt, max_length=self.tokenizer_max_length + drop_idx, padding=True, truncation=True, return_tensors="pt"
204 ).to(device)
--> 205 encoder_hidden_states = self.text_encoder(
206 input_ids=txt_tokens.input_ids,
207 attention_mask=txt_tokens.attention_mask,
/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py in _wrapped_call_impl(self, *args, **kwargs)
1737 return self._compiled_call_impl(*args, **kwargs) # type: ignore[misc]
1738 else:
-> 1739 return self._call_impl(*args, **kwargs)
1740
1741 # torchrec tests the code consistency with the following code
/usr/local/lib/python3.11/dist-packages/torch/nn/modules/module.py in _call_impl(self, *args, **kwargs)
1748 or _global_backward_pre_hooks or _global_backward_hooks
1749 or _global_forward_hooks or _global_forward_pre_hooks):
-> 1750 return forward_call(*args, **kwargs)
1751
1752 result = None
/usr/local/lib/python3.11/dist-packages/accelerate/hooks.py in new_forward(module, *args, **kwargs)
173 output = module._old_forward(*args, **kwargs)
174 else:
--> 175 output = module._old_forward(*args, **kwargs)
176 return module._hf_hook.post_forward(module, output)
177
/usr/local/lib/python3.11/dist-packages/transformers/utils/generic.py in wrapper(self, *args, **kwargs)
941
942 try:
--> 943 output = func(self, *args, **kwargs)
944 if is_requested_to_return_tuple or (is_configured_to_return_tuple and is_top_level_module):
945 | https://github.com/huggingface/diffusers/issues/12719 | open | [] | 2025-11-26T08:35:46Z | 2025-11-26T09:15:54Z | null | chaowenguo |
vllm-project/vllm | 29,474 | [P/D][Metrics] Consider combined/summed metrics (e.g. ttft and e2e_request_latency) for prefill and decode instances | ### Your current environment
<details>
<summary>Env info snipped</summary>
```
Collecting environment information...
uv is set
==============================
System Info
==============================
OS : Ubuntu 24.04.1 LTS (x86_64)
GCC version : (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0
Clang version : Could not collect
CMake version : version 3.28.3
Libc version : glibc-2.39
==============================
PyTorch Info
==============================
PyTorch version : 2.8.0+cu128
Is debug build : False
CUDA used to build PyTorch : 12.8
ROCM used to build PyTorch : N/A
==============================
Python Environment
==============================
Python version : 3.12.3 (main, Aug 14 2025, 17:47:21) [GCC 13.3.0] (64-bit runtime)
Python platform : Linux-5.15.0-152-generic-x86_64-with-glibc2.39
==============================
CUDA / GPU Info
==============================
Is CUDA available : True
CUDA runtime version : 12.8.93
CUDA_MODULE_LOADING set to : LAZY
GPU models and configuration :
GPU 0: NVIDIA H20
GPU 1: NVIDIA H20
GPU 2: NVIDIA H20
GPU 3: NVIDIA H20
GPU 4: NVIDIA H20
GPU 5: NVIDIA H20
GPU 6: NVIDIA H20
GPU 7: NVIDIA H20
Nvidia driver version : 570.172.08
cuDNN version : Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.9.8.0
/usr/lib/x86_64-linux-gnu/libcudnn_adv.so.9.8.0
/usr/lib/x86_64-linux-gnu/libcudnn_cnn.so.9.8.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_precompiled.so.9.8.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_runtime_compiled.so.9.8.0
/usr/lib/x86_64-linux-gnu/libcudnn_graph.so.9.8.0
/usr/lib/x86_64-linux-gnu/libcudnn_heuristic.so.9.8.0
/usr/lib/x86_64-linux-gnu/libcudnn_ops.so.9.8.0
HIP runtime version : N/A
MIOpen runtime version : N/A
Is XNNPACK available : True
==============================
CPU Info
==============================
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 46 bits physical, 57 bits virtual
Byte Order: Little Endian
CPU(s): 128
On-line CPU(s) list: 0-127
Vendor ID: GenuineIntel
BIOS Vendor ID: Intel(R) Corporation
Model name: INTEL(R) XEON(R) PLATINUM 8562Y+
BIOS Model name: INTEL(R) XEON(R) PLATINUM 8562Y+ CPU @ 2.8GHz
BIOS CPU family: 179
CPU family: 6
Model: 207
Thread(s) per core: 2
Core(s) per socket: 32
Socket(s): 2
Stepping: 2
CPU(s) scaling MHz: 73%
CPU max MHz: 4100.0000
CPU min MHz: 800.0000
BogoMIPS: 5600.00
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 invpcid_single intel_ppin cdp_l2 ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid bus_lock_detect cldemote movdiri movdir64b enqcmd fsrm md_clear serialize tsxldtrk pconfig arch_lbr amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
Virtualization: VT-x
L1d cache: 3 MiB (64 instances)
L1i cache: 2 MiB (64 instances)
L2 cache: 128 MiB (64 instances)
L3 cache: 120 MiB (2 instances)
NUMA node(s): 2
NUMA node0 CPU(s): 0-31,64-95
NUMA node1 CPU(s): 32-63,96-127
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target s | https://github.com/vllm-project/vllm/issues/29474 | open | [
"usage",
"kv-connector"
] | 2025-11-26T02:50:17Z | 2025-11-26T08:31:18Z | 1 | mgw2168-1 |
vllm-project/vllm | 29,472 | [Installation]: how to Install vllm on dell promax gb10 | ### Your current environment
I failed to install vllm on dell promax gb10 , mesages as followed
nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2025 NVIDIA Corporation
Built on Wed_Aug_20_01:57:39_PM_PDT_2025
Cuda compilation tools, release 13.0, V13.0.88
Build cuda_13.0.r13.0/compiler.36424714_0
pip install vllm
Successfully installed torch-2.9.0 torchaudio-2.9.0 torchvision-0.24.0 vllm-0.11.2
```
(py312) dell@promaxgb10-0843:~/test/vllm/Qwen$ vllm -V
Traceback (most recent call last):
File "/home/dell/miniconda3/envs/py312/bin/vllm", line 3, in <module>
from vllm.entrypoints.cli.main import main
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/entrypoints/cli/__init__.py", line 3, in <module>
from vllm.entrypoints.cli.benchmark.latency import BenchmarkLatencySubcommand
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/entrypoints/cli/benchmark/latency.py", line 5, in <module>
from vllm.benchmarks.latency import add_cli_args, main
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/benchmarks/latency.py", line 17, in <module>
from vllm.engine.arg_utils import EngineArgs
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/engine/arg_utils.py", line 35, in <module>
from vllm.attention.backends.registry import AttentionBackendEnum
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/attention/__init__.py", line 4, in <module>
from vllm.attention.backends.abstract import (
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/attention/backends/abstract.py", line 9, in <module>
from vllm.model_executor.layers.linear import ColumnParallelLinear
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/model_executor/__init__.py", line 4, in <module>
from vllm.model_executor.parameter import BasevLLMParameter, PackedvLLMParameter
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/model_executor/parameter.py", line 11, in <module>
from vllm.distributed import (
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/distributed/__init__.py", line 4, in <module>
from .communication_op import *
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/distributed/communication_op.py", line 9, in <module>
from .parallel_state import get_tp_group
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/distributed/parallel_state.py", line 250, in <module>
direct_register_custom_op(
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/utils/torch_utils.py", line 640, in direct_register_custom_op
from vllm.platforms import current_platform
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/platforms/__init__.py", line 257, in __getattr__
_current_platform = resolve_obj_by_qualname(platform_cls_qualname)()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/utils/import_utils.py", line 89, in resolve_obj_by_qualname
module = importlib.import_module(module_name)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/dell/miniconda3/envs/py312/lib/python3.12/importlib/__init__.py", line 90, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/dell/miniconda3/envs/py312/lib/python3.12/site-packages/vllm/platforms/cuda.py", line 16, in <module>
import vllm._C # noqa
^^^^^^^^^^^^^^
ImportError: libtorch_cuda.so: cannot open shared object file: No such file or directory
```
### How you are installing vllm
```sh
pip install vllm
```
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29472 | open | [
"installation"
] | 2025-11-26T02:41:18Z | 2026-01-01T12:28:29Z | 2 | goactiongo |
vllm-project/vllm | 29,436 | [Bug]: vLLM Serve with LMCache enabled produces wrong output for GPT-OSS-20B | ### Your current environment
<details>
<summary>The output of <code>python collect_env.py</code></summary>
```text
Your output of `python collect_env.py` here
```
</details>
### 🐛 Describe the bug
vLLM serve command with LMCache enabled produces wrong output with GPT OSS 20B for subsequent invocations with the same prompt
Steps to reproduce:
Command to start the server:
```
LMCACHE_CONFIG_FILE=lmcache_cpu.yaml
vllm serve openai/gpt-oss-20b --port 8000 --kv-transfer-config '{"kv_connector":"LMCacheConnectorV1", "kv_role":"kv_both"}'
```
Invocation:
```
curl 127.0.0.1:8000/v1/chat/completions -H "Content-Type: application/json" -d '{"model": "openai/gpt-oss-20b", "messages": [ {"role": "user", "content": "What is Amazon SageMaker?"}]}'
```
First invocation:
```
{
"id":"chatcmpl-951ca7178b1e4226b0343cb070033487",
"object":"chat.completion",
"created":1764098087,
"model":"openai/gpt-oss-20b",
"choices":[
{"index":0,"message":{"role":"assistant","content":"**Amazon SageMaker** is Amazon Web Services’ fully‑managed platform that lets you build, train, tune, and deploy machine‑learning models fast—without managing the underlying infrastructure.\n\nKey capabilities\n\n| Feature | What it does |\n|--------|--------------|\n| **SageMaker Studio** | A web‑based IDE that bundles notebooks, visual debugging, model monitoring, and collaboration tools. |\n| **Built‑in algorithms & frameworks** | Pre‑packaged models (XGBoost, Linear Learner, etc.) and support for your own TensorFlow, PyTorch, MXNet, Scikit‑learn, R, etc. |\n| **Auto‑ML & automated model tuning** | SageMaker Autopilot automatically searches model architectures and hyper‑parameters. |\n| **Managed training** | Spot, distributed, and GPU training jobs that scale to the required compute. |\n| **Model deployment** | One‑click production endpoints, batch transform, edge inference (SageMaker Edge), and real‑time or asynchronous inference. |\n| **Inference pipelines** | Compose multiple models or processing steps into a single pipeline. |\n| **Model monitoring & A/B testing** | Continuous evaluation of drift, predictions, and performance metrics. |\n| **Security & compliance** | VPC, IAM, KMS encryption, private cataloging, and audit trails. |\n\nIn short, SageMaker removes the operational burden of ML—so teams can focus on data science and business value rather than servers, networking, and scaling.","refusal":null,"annotations":null,"audio":null,"function_call":null,"tool_calls":[],"reasoning":"User asks \"What is Amazon SageMaker?\" Short answer. Provide description: fully managed ML service, environment to build, train, deploy models, etc. Should be succinct.","reasoning_content":"User asks \"What is Amazon SageMaker?\" Short answer. Provide description: fully managed ML service, environment to build, train, deploy models, etc. Should be succinct."},"logprobs":null,"finish_reason":"stop","stop_reason":null,"token_ids":null}],"service_tier":null,"system_fingerprint":null,"usage":{"prompt_tokens":75,"total_tokens":426,"completion_tokens":351,"prompt_tokens_details":null},"prompt_logprobs":null,"prompt_token_ids":null,"kv_transfer_params":null}
```
Second invocation:
```
{
"id": "chatcmpl-4ebc19fc5c2a41a7bebc01ea8d1c98b1",
"object": "chat.completion",
"created": 1764098160,
"model": "openai/gpt-oss-20b",
"choices": [
{
"index": 0,
"message": {
"role": "assistant",
"content": "Sure! Here’s a basic guide to get you started with writing a cool, informative yet accessible article on **\"The Fascinating World of Quantum Computing\"** for a general audience. Feel free to adapt the structure, tone, or content to match your style and publication’s guidelines.\n\n---\n\n## 1. Hook & Context (≈150–200 words)\n\n- **Start with a vivid anecdote, surprising fact, or a relatable analogy** that introduces the “wow” moment in quantum computing.\n - *Example:* “Imagine a coin that, instead of being heads or tails, can be both at the same time… until you look at it.” \n- **Briefly state why this topic matters** to everyday life: faster drug discovery, better encryption, breakthrough materials, etc.\n\n> **Tell readers what they’ll learn**: a quick glimpse of quantum fundamentals, why it’s different from classic bits, and how it could reshape technologies.\n\n---\n\n## 2. What’s a Quantum Computer? (≈300 words)\n\n| Section | Content | Quick Tips |\n|---------|---------|------------|\n| **2.1 “Bits” vs. “Qubits”** | • Classical bits (“0” or “1”).<br>• Qubits: superposition (both 0 & 1) & entanglement. | Use visual metaphors: a spinning top (superposition) and two dancers always in sync (entanglement). |\n| **2.2 Basic Operations** | • Quantum gates (Pauli X, H, CNOT).<br>• The role of interference. | A tiny “reversible” logic of the quantum “if‑then” that flips outcomes. |\n| **2.3 Measuring As a Collapses** | • Outcome collapse on measurement.<br>• Probabilities & expectation values. | Compare to a gamble: you only learn the re | https://github.com/vllm-project/vllm/issues/29436 | open | [
"bug"
] | 2025-11-25T19:27:24Z | 2025-11-25T19:27:24Z | 0 | ksuma2109 |
vllm-project/vllm | 29,409 | [Usage]: Custom Logits Processors V1 how to get tokenizer into processor | ### Problem with tokenizer
For the second day now, I've been unable to figure out how to get a tokenizer inside a custom processor. I used the processor from the documentation as an example. I examined each object through debug, but couldn't find where to extract the tokenizer. In v0, this was done simply at the request level, by passing an argument to the object.
How to pass a tokenizer to the processor?
```python import torch
from vllm.config import VllmConfig
from vllm.sampling_params import SamplingParams
from vllm.v1.sample.logits_processor import (BatchUpdate,
LogitsProcessor,
MoveDirectionality)
class DummyLogitsProcessor(LogitsProcessor):
"""Fake logit processor to support unit testing and examples"""
@classmethod
def validate_params(cls, params: SamplingParams):
target_token: int | None = params.extra_args and params.extra_args.get(
"target_token"
)
if target_token is not None and not isinstance(target_token, int):
raise ValueError(f"target_token value {target_token} is not int")
def __init__(self, vllm_config: "VllmConfig", device: torch.device,
is_pin_memory: bool):
self.req_info: dict[int, int] = {}
def is_argmax_invariant(self) -> bool:
"""Never impacts greedy sampling"""
return False
def update_state(self, batch_update: BatchUpdate | None):
if not batch_update:
return
# Process added requests.
for index, params, _, _ in batch_update.added:
assert params is not None
self.validate_params(params)
if params.extra_args and (target_token :=
params.extra_args.get("target_token")):
self.req_info[index] = target_token
else:
self.req_info.pop(index, None)
if self.req_info:
# Process removed requests.
for index in batch_update.removed:
self.req_info.pop(index, None)
# Process moved requests, unidirectional move (a->b) and swap
# (a<->b)
for adx, bdx, direct in batch_update.moved:
a_val = self.req_info.pop(adx, None)
b_val = self.req_info.pop(bdx, None)
if a_val is not None:
self.req_info[bdx] = a_val
if direct == MoveDirectionality.SWAP and b_val is not None:
self.req_info[adx] = b_val
def apply(self, logits: torch.Tensor) -> torch.Tensor:
if not self.req_info:
return logits
# Save target values before modification
cols = torch.tensor(
list(self.req_info.values()), dtype=torch.long, device=logits.device
)
rows = torch.tensor(
list(self.req_info.keys()), dtype=torch.long, device=logits.device
)
values_to_keep = logits[rows, cols].clone()
# Mask all but target tokens
logits[rows] = float('-inf')
logits[rows, cols] = values_to_keep
return logits
```
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29409 | closed | [
"usage"
] | 2025-11-25T13:24:17Z | 2025-12-02T10:33:18Z | 6 | cvadim130 |
vllm-project/vllm | 29,389 | [Bug]: race condition in shm_broadcast.py | ### Your current environment
<details>
<summary>The output of <code>python collect_env.py</code></summary>
```text
Your output of `python collect_env.py` here
```
</details>
### 🐛 Describe the bug
# Problem
`ShmRingBuffer` is a lock-free queue, the implementation of which https://github.com/vllm-project/vllm/blob/12c007e288bf5c0ae3bd438036fbafbad88e706b/vllm/distributed/device_communicators/shm_broadcast.py#L98-L153
relies on the fact that when a flag is written to, signalling a valid state, the associated data is also in a valid state. To illustrate the point, consider the program
```python
shm = shared_memory.SharedMemory(..., size=128)
# set shm to 0
# process 1
shm[0] = 1
shm[64] = 1
# process 2
while shm[64] != 1:
pass
print(shm[0])
```
`ShmRingBuffer` requires that `print(shm[0])` always prints `1`. **There is no guarantee this is true**. For this to be true,
1. The Python language/implementation must provide a memory model, which it doesn't. Loosely speaking, a memory model is a set of guarantees on how source code maps to hardware instructions.
2. Even if we assume the source code maps "as intended" to hardware instructions, the hardware must ensure that process 2 must observe the writes to `shm[0]` and `shm[64]` in the same order as process 1.
An example of 2 breaking down is given in [`race_condition.cpp`](https://gist.github.com/nvjullin/cc52386e291fe41218b54406ece962a0). On an ARM CPU,
```bash
$ g++ -std=c++17 race_condition.cpp
$ ./a.out
number of violations: 5
# ...
```
Unfortunately, I don't know how to demonstrate the same race condition in Python.
# What it means
`ShmRingBuffer` can have corrupted memory and crashes vLLM sporadically. Such a crash would be near impossible to reproduce and debug.
# Solutions
In order of recommendation:
1. Remove `ShmRingBuffer` and always use the fallback `self.local_socket.send(serialized_obj)`. This is the simplest.
2. Use a well-tested lock-free queue implementation and don't write our own. Lock-free programming is notoriously difficult to write correctly, requires expertise to understand and is overall a maintenence nightmare.
3. Write it in C++ with proper atomics that guarantees the ordering of writes. The implementation should document extensively the proof of its correctness across different architectures. Python provides no tools for lock-free programming, making it impossible to write.
CC @youkaichao @nvpohanh
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29389 | open | [
"bug"
] | 2025-11-25T09:25:52Z | 2025-11-25T09:25:52Z | 0 | nvjullin |
vllm-project/vllm | 29,382 | [Doc]: Expert Parallel Deployment says "Tensor parallel size (always 1 for now)" is confusing | ### 📚 The doc issue
On page https://docs.vllm.ai/en/latest/serving/expert_parallel_deployment/#single-node-deployment it says Tensor parallel size can only be 1 but didn't mention the behavior of Attention Layers
On page https://docs.vllm.ai/en/latest/serving/data_parallel_deployment/ it says The expert layers will by default form a (DP x TP) sized tensor parallel group. To enable expert parallelism, include the --enable-expert-parallel CLI arg (on all nodes in the multi-node case).
which is rather confusing.
### Suggest a potential alternative/fix
Point out the correct behavior of MoE models when TP, EP are both set.
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29382 | closed | [
"documentation"
] | 2025-11-25T07:54:42Z | 2025-12-13T17:38:01Z | 0 | xeonliu |
huggingface/transformers | 42,375 | SAM3 single image inference with multiple text prompt | Hi
I'm trying to run inference on a single image, aiming to get the bbox of objects from several different categories (e.g. "a person" and "a car").
the only example i found for prompting with multiple categories is in the "Batched Inference with Text Prompts" example, but then i need to unnecessarily duplicate my image as the # of categories.
is there a different more efficient way of achieving this?
p.s
when i try prompting with a list of several categories and a single image i get an error.
| https://github.com/huggingface/transformers/issues/42375 | open | [] | 2025-11-25T06:20:09Z | 2026-01-05T16:16:01Z | 9 | iariav |
huggingface/trl | 4,569 | [doc issue] doc on "GRPO with replay buffer" buggy | ### Reproduction
The code example in [doc for "GRPO with replay buffer"](https://huggingface.co/docs/trl/main/en/experimental#grpo-with-replay-buffer) is kind of buggy.
- It imports `GRPOWithReplayBufferTrainer` but never used.
- It uses `GRPOWithReplayBufferConfig` but never imported
- The code is apparently not executable.
Below is the code example given in the doc:
```python
from trl.experimental.grpo_with_replay_buffer import GRPOWithReplayBufferTrainer
from datasets import load_dataset
dataset = load_dataset("trl-internal-testing/zen", "standard_prompt_only", split="train")
# Guarantee that some rewards have 0 std
def custom_reward_func(completions, **kwargs):
if torch.rand(1).item() < 0.25:
return [0] * len(completions) # simulate some None rewards
else:
return torch.rand(len(completions)).tolist()
training_args = GRPOWithReplayBufferConfig(
output_dir=self.tmp_dir,
learning_rate=1e-4,
per_device_train_batch_size=4,
num_generations=4,
max_completion_length=8,
replay_buffer_size=8,
report_to="none",
)
trainer = GRPOTrainer(
model="trl-internal-testing/tiny-Qwen2ForCausalLM-2.5",
reward_funcs=[custom_reward_func],
args=training_args,
train_dataset=dataset,
)
previous_trainable_params = {n: param.clone() for n, param in trainer.model.named_parameters()}
trainer.train()
```
### System Info
NA
### Checklist
- [x] I have checked that my issue isn't already filed (see [open issues](https://github.com/huggingface/trl/issues?q=is%3Aissue))
- [x] I have included my system information
- [x] Any code provided is minimal, complete, and reproducible ([more on MREs](https://docs.github.com/en/get-started/writing-on-github/working-with-advanced-formatting/creating-and-highlighting-code-blocks))
- [x] Any code provided is properly formatted in code blocks, (no screenshot, [more on code blocks](https://docs.github.com/en/get-started/writing-on-github/working-with-advanced-formatting/creating-and-highlighting-code-blocks))
- [x] Any traceback provided is complete | https://github.com/huggingface/trl/issues/4569 | closed | [
"🐛 bug",
"📚 documentation",
"🏋 GRPO"
] | 2025-11-25T01:30:28Z | 2025-11-25T21:28:00Z | 2 | DNXie |
vllm-project/vllm | 29,306 | [Usage]: dots.llm.inst is not running due to a type error | ### Your current environment
I'm trying to run dots llm on 4xH100
```
vllm serve \
--uvicorn-log-level=info \
rednote-hilab/dots.llm1.inst \
--dtype auto \
--api-key xxx \
--host 0.0.0.0 \
--port 8000 \
--tensor-parallel-size 4
--ipc=host \
--trust-remote-code
```
It failed to run, I got the following crash:
```text
(EngineCore_DP0 pid=10684) ERROR 11-24 09:41:25 [v1/executor/multiproc_executor.py:230] Worker proc VllmWorker-1 died unexpectedly, shutting down executor.
(EngineCore_DP0 pid=10684) Process EngineCore_DP0:
(EngineCore_DP0 pid=10684) Traceback (most recent call last):
(EngineCore_DP0 pid=10684) File "/usr/lib/python3.12/multiprocessing/process.py", line 314, in _bootstrap
(EngineCore_DP0 pid=10684) self.run()
(EngineCore_DP0 pid=10684) File "/usr/lib/python3.12/multiprocessing/process.py", line 108, in run
(EngineCore_DP0 pid=10684) self._target(*self._args, **self._kwargs)
(EngineCore_DP0 pid=10684) File "/home/ubuntu/venv/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 846, in run_engine_core
(EngineCore_DP0 pid=10684) raise e
(EngineCore_DP0 pid=10684) File "/home/ubuntu/venv/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 833, in run_engine_core
(EngineCore_DP0 pid=10684) engine_core = EngineCoreProc(*args, **kwargs)
(EngineCore_DP0 pid=10684) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore_DP0 pid=10684) File "/home/ubuntu/venv/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 606, in __init__
(EngineCore_DP0 pid=10684) super().__init__(
(EngineCore_DP0 pid=10684) File "/home/ubuntu/venv/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 109, in __init__
(EngineCore_DP0 pid=10684) num_gpu_blocks, num_cpu_blocks, kv_cache_config = self._initialize_kv_caches(
(EngineCore_DP0 pid=10684) ^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore_DP0 pid=10684) File "/home/ubuntu/venv/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 231, in _initialize_kv_caches
(EngineCore_DP0 pid=10684) available_gpu_memory = self.model_executor.determine_available_memory()
(EngineCore_DP0 pid=10684) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore_DP0 pid=10684) File "/home/ubuntu/venv/lib/python3.12/site-packages/vllm/v1/executor/abstract.py", line 126, in determine_available_memory
(EngineCore_DP0 pid=10684) return self.collective_rpc("determine_available_memory")
(EngineCore_DP0 pid=10684) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore_DP0 pid=10684) File "/home/ubuntu/venv/lib/python3.12/site-packages/vllm/v1/executor/multiproc_executor.py", line 358, in collective_rpc
(EngineCore_DP0 pid=10684) return aggregate(get_response())
(EngineCore_DP0 pid=10684) ^^^^^^^^^^^^^^
(EngineCore_DP0 pid=10684) File "/home/ubuntu/venv/lib/python3.12/site-packages/vllm/v1/executor/multiproc_executor.py", line 341, in get_response
(EngineCore_DP0 pid=10684) raise RuntimeError(
(EngineCore_DP0 pid=10684) RuntimeError: Worker failed with error 'TypeError: can't multiply sequence by non-int of type 'float'
(EngineCore_DP0 pid=11385) ERROR 11-24 09:45:27 [v1/engine/core.py:842] EngineCore failed to start.
(EngineCore_DP0 pid=11385) ERROR 11-24 09:45:27 [v1/engine/core.py:842] Traceback (most recent call last):
(EngineCore_DP0 pid=11385) ERROR 11-24 09:45:27 [v1/engine/core.py:842] File "/home/ubuntu/venv/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 833, in run_engine_core
(EngineCore_DP0 pid=11385) ERROR 11-24 09:45:27 [v1/engine/core.py:842] engine_core = EngineCoreProc(*args, **kwargs)
(EngineCore_DP0 pid=11385) ERROR 11-24 09:45:27 [v1/engine/core.py:842] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore_DP0 pid=11385) ERROR 11-24 09:45:27 [v1/engine/core.py:842] File "/home/ubuntu/venv/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 606, in __init__
(EngineCore_DP0 pid=11385) ERROR 11-24 09:45:27 [v1/engine/core.py:842] super().__init__(
(EngineCore_DP0 pid=11385) ERROR 11-24 09:45:27 [v1/engine/core.py:842] File "/home/ubuntu/venv/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 109, in __init__
(EngineCore_DP0 pid=11385) ERROR 11-24 09:45:27 [v1/engine/core.py:842] num_gpu_blocks, num_cpu_blocks, kv_cache_config = self._initialize_kv_caches(
(EngineCore_DP0 pid=11385) ERROR 11-24 09:45:27 [v1/engine/core.py:842] ^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore_DP0 pid=11385) ERROR 11-24 09:45:27 [v1/engine/core.py:842] File "/home/ubuntu/venv/lib/python3.12/site-packages/vllm/v1/engine/core.py", line 231, in _initialize_kv_caches
(EngineCore_DP0 pid=11385) ERROR 11-24 09:45:27 [v1/engine/core.py:842] available_gpu_memory = self.model_executor.determine_available_memory()
(EngineCore_DP0 pid=11385) ERROR 11-24 09:45:27 [v1/engine/core.py:842] | https://github.com/vllm-project/vllm/issues/29306 | closed | [
"usage"
] | 2025-11-24T09:48:08Z | 2025-11-28T23:25:27Z | 1 | rain-1 |
huggingface/transformers | 42,353 | SAM3 point mode is not supported yet? | In [SAM3 official example](https://github.com/facebookresearch/sam3/blob/main/examples/sam3_for_sam1_task_example.ipynb
), they also support point mode. But it seems that transforms has not supported yet?
| https://github.com/huggingface/transformers/issues/42353 | closed | [] | 2025-11-24T07:16:52Z | 2025-11-26T15:16:25Z | 1 | haofanwang |
vllm-project/vllm | 29,297 | [Bug]: What should the image embedding input be like? I have tested with multiple cases but it all fails | ### Your current environment
```text
==============================
System Info
==============================
OS : Red Hat Enterprise Linux release 8.10 (Ootpa) (x86_64)
GCC version : (GCC) 8.5.0 20210514 (Red Hat 8.5.0-26)
Clang version : Could not collect
CMake version : Could not collect
Libc version : glibc-2.28
==============================
PyTorch Info
==============================
PyTorch version : 2.9.0+cu128
Is debug build : False
CUDA used to build PyTorch : 12.8
ROCM used to build PyTorch : N/A
==============================
Python Environment
==============================
Python version : 3.10.19 (main, Oct 21 2025, 16:43:05) [GCC 11.2.0] (64-bit runtime)
Python platform : Linux-4.18.0-553.50.1.el8_10.x86_64-x86_64-with-glibc2.28
==============================
CUDA / GPU Info
==============================
Is CUDA available : True
CUDA runtime version : Could not collect
CUDA_MODULE_LOADING set to :
GPU models and configuration :
GPU 0: NVIDIA A100-SXM4-40GB
GPU 1: NVIDIA A100-SXM4-40GB
Nvidia driver version : 575.51.03
cuDNN version : Could not collect
HIP runtime version : N/A
MIOpen runtime version : N/A
Is XNNPACK available : True
==============================
CPU Info
==============================
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
CPU(s): 128
On-line CPU(s) list: 0-127
Thread(s) per core: 1
Core(s) per socket: 64
Socket(s): 2
NUMA node(s): 8
Vendor ID: AuthenticAMD
CPU family: 23
Model: 49
Model name: AMD EPYC 7742 64-Core Processor
Stepping: 0
CPU MHz: 2250.000
CPU max MHz: 2250.0000
CPU min MHz: 1500.0000
BogoMIPS: 4491.72
Virtualization: AMD-V
L1d cache: 32K
L1i cache: 32K
L2 cache: 512K
L3 cache: 16384K
NUMA node0 CPU(s): 0-15
NUMA node1 CPU(s): 16-31
NUMA node2 CPU(s): 32-47
NUMA node3 CPU(s): 48-63
NUMA node4 CPU(s): 64-79
NUMA node5 CPU(s): 80-95
NUMA node6 CPU(s): 96-111
NUMA node7 CPU(s): 112-127
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid aperfmperf pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 hw_pstate ssbd mba ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 cqm rdt_a rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local clzero irperf xsaveerptr wbnoinvd arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif v_spec_ctrl umip rdpid overflow_recov succor smca sme sev sev_es
==============================
Versions of relevant libraries
==============================
[pip3] flashinfer-python==0.5.2
[pip3] numpy==2.2.6
[pip3] nvidia-cublas-cu12==12.8.4.1
[pip3] nvidia-cuda-cupti-cu12==12.8.90
[pip3] nvidia-cuda-nvrtc-cu12==12.8.93
[pip3] nvidia-cuda-runtime-cu12==12.8.90
[pip3] nvidia-cudnn-cu12==9.10.2.21
[pip3] nvidia-cudnn-frontend==1.16.0
[pip3] nvidia-cufft-cu12==11.3.3.83
[pip3] nvidia-cufile-cu12==1.13.1.3
[pip3] nvidia-curand-cu12==10.3.9.90
[pip3] nvidia-cusolver-cu12==11.7.3.90
[pip3] nvidia-cusparse-cu12==12.5.8.93
[pip3] nvidia-cusparselt-cu12==0.7.1
[pip3] nvidia-cutlass-dsl==4.3.0
[pip3] nvidia-ml-py==13.580.82
[pip3] nvidia-nccl-cu12==2.27.5
[pip3] nvidia-nvjitlink-cu12==12.8.93
[pip3] nvidia-nvshmem-cu12==3.3.20
[pip3] nvidia-nvtx-cu12==12.8.90
[pip3] pyzmq==27.1.0
[pip3] torch==2.9.0
[pip3] torchaudio==2.9.0
[pip3] torchvision==0.24.0
[pip3] transformers==4.57.1
[pip3] triton==3.5.0
[conda] flashinfer-python 0.5.2 pypi_0 pypi
[conda] numpy 2.2.6 pypi_0 pypi
[conda] nvidia-cublas-cu12 12.8.4.1 pypi_0 pypi
[conda] nvidia-cuda-cupti-cu12 12.8.90 pypi_0 pypi
[conda] nvidia-cuda-nvrtc-cu12 12.8.93 pypi_0 pypi
[conda] nvidia-cuda-runtime-cu12 12.8.90 pypi_0 pypi
[conda] nvidia-cudnn-cu12 9.10.2.21 pypi_0 pypi
[conda] nvidia-cudn | https://github.com/vllm-project/vllm/issues/29297 | closed | [
"usage"
] | 2025-11-24T06:02:09Z | 2025-11-26T13:00:17Z | 2 | DamonZhao-sfu |
vllm-project/vllm | 29,294 | [CPU Backend] [Doc]: Update Installation Docs for Arm CPUs | ### 📚 The doc issue
This page https://docs.vllm.ai/en/stable/getting_started/installation/cpu/#arm-aarch64 is very out-dated.
We now release Arm CPU wheels and images thanks to #26931 and #27331
We need to update that page to reflect that :)
### Suggest a potential alternative/fix
_No response_
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29294 | closed | [
"documentation",
"cpu"
] | 2025-11-24T05:33:46Z | 2025-12-15T19:46:26Z | 5 | fadara01 |
vllm-project/vllm | 29,286 | [Performance]: cache system prompt token ids | ### Proposal to improve performance
As system prompt can be very long now, tokenize the system prompt can be slow.
Using H20, tokenize 5000 tokens cost about 10ms as below:
System prompts are usually fixed and reusable, so cache the system prompt can be profitable.
Specificly:
1. In **apply_hf_chat_template** method we can separate the system prompt from other prompts, we can use condition **cache_system_prompt = truncate_prompt_tokens is None and not tokenize and len(conversation) > 1 and conversation[0].get("role") == "system"** to judge when we should separate the system prompt.
2. In **_normalize_prompt_text_to_input** method we judge that whether system prompt is in the dict ({system prompt: token ids}) that we can reuse, then concat system prompt token ids and prompt token ids as the final input_ids.
I am willing to contribute to this opt and looking forward to your suggestions!
### Report of performance regression
The above cost can be profitable.
### Misc discussion on performance
_No response_
### Your current environment (if you think it is necessary)
```text
The output of `python collect_env.py`
```
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29286 | open | [
"performance"
] | 2025-11-24T01:55:32Z | 2025-11-28T08:57:06Z | 2 | Eviannn |
vllm-project/vllm | 29,281 | [Usage]: Removing last generated token from output and kv cache | ### Your current environment
```text
vLLM 0.11.2
```
### How would you like to use vllm
Hey guys,
i am currently working on a research project where i load a moe-like model and i want to do routing based on the sequence state.
The goal is to let expert 0 generate until it reaches the eos token, then remove the eos token and finish generation with expert 1 until the eos token is hit a second time.
I want to do this to use different strengths of both models.
My current approach is to modify GPUModelRunner and Scheduler to remove the eos token from output, reduce num_computed_tokens by 1 and compute a static routing tensor based on the sequence state which i pass as additional model input, to route to expert 0 or 1.
Now i am having some issues with unexpected output, especially with tensor_parallelism>1 on multiple gpus.
I was wondering if there already is a reliable solution to remove the last generated token from output and kv cache, so that the computation leading to eos does not interfere with the second expert.
Or maybe there is even a better way to do this?
Thank you!
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29281 | closed | [
"usage"
] | 2025-11-23T22:39:16Z | 2025-11-26T09:33:53Z | 0 | josefdra |
vllm-project/vllm | 29,277 | [Usage]: Creating and accessing per request arguments inside vLLM model | ### Your current environment
```text
The output of `python collect_env.py`
```
### How would you like to use vllm
I want to implement token compression techniques on the output embeddings of Qwen-2.5VL which would occur dynamically as the number of requests change. Is there anyway to implement this in vLLM? I see that SamplingParams seem to be the only way to use per request custom arguments but I don’t believe it can be accessed within the model code directly?
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29277 | open | [
"usage"
] | 2025-11-23T21:59:31Z | 2025-11-23T21:59:31Z | 0 | minlu21 |
huggingface/transformers | 42,344 | How to fine-tune SAM 3D models? | ### Model description
The recently released SAM 3D work is truly remarkable. Do you plan to integrate it into Transformers and enable fine-tuning?
https://huggingface.co/facebook/sam-3d-objects
### Open source status
- [x] The model implementation is available
- [x] The model weights are available
### Provide useful links for the implementation
_No response_ | https://github.com/huggingface/transformers/issues/42344 | open | [
"New model"
] | 2025-11-23T17:40:57Z | 2025-11-23T17:40:57Z | null | bruno686 |
vllm-project/vllm | 29,264 | [Usage]: Monkey Patching SamplingParams | ### Your current environment
```text
Collecting environment information...
==============================
System Info
==============================
OS : Ubuntu 24.04.3 LTS (x86_64)
GCC version : (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0
Clang version : Could not collect
CMake version : version 3.28.3
Libc version : glibc-2.39
==============================
PyTorch Info
==============================
PyTorch version : 2.9.0+cu128
Is debug build : False
CUDA used to build PyTorch : 12.8
ROCM used to build PyTorch : N/A
==============================
Python Environment
==============================
Python version : 3.13.5 | packaged by conda-forge | (main, Jun 16 2025, 08:27:50) [GCC 13.3.0] (64-bit runtime)
Python platform : Linux-6.8.0-87-generic-x86_64-with-glibc2.39
==============================
CUDA / GPU Info
==============================
Is CUDA available : True
CUDA runtime version : 12.8.93
CUDA_MODULE_LOADING set to :
GPU models and configuration :
GPU 0: NVIDIA B200
GPU 1: NVIDIA B200
GPU 2: NVIDIA B200
GPU 3: NVIDIA B200
GPU 4: NVIDIA B200
GPU 5: NVIDIA B200
GPU 6: NVIDIA B200
GPU 7: NVIDIA B200
Nvidia driver version : 570.195.03
cuDNN version : Could not collect
HIP runtime version : N/A
MIOpen runtime version : N/A
Is XNNPACK available : True
==============================
CPU Info
==============================
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 52 bits physical, 57 bits virtual
Byte Order: Little Endian
CPU(s): 224
On-line CPU(s) list: 0-223
Vendor ID: GenuineIntel
Model name: INTEL(R) XEON(R) PLATINUM 8570
CPU family: 6
Model: 207
Thread(s) per core: 2
Core(s) per socket: 56
Socket(s): 2
Stepping: 2
CPU(s) scaling MHz: 31%
CPU max MHz: 4000.0000
CPU min MHz: 800.0000
BogoMIPS: 4200.00
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 intel_ppin cdp_l2 ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect user_shstk avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req vnmi avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid bus_lock_detect cldemote movdiri movdir64b enqcmd fsrm md_clear serialize tsxldtrk pconfig arch_lbr ibt amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities ibpb_exit_to_user
Virtualization: VT-x
L1d cache: 5.3 MiB (112 instances)
L1i cache: 3.5 MiB (112 instances)
L2 cache: 224 MiB (112 instances)
L3 cache: 600 MiB (2 instances)
NUMA node(s): 2
NUMA node0 CPU(s): 0-55,112-167
NUMA node1 CPU(s): 56-111,168-223
Vulnerability Gather data sampling: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Enhanced / Automatic IBRS; IBPB conditional; RSB filling; PBRSB-eIBRS SW sequence; BHI | https://github.com/vllm-project/vllm/issues/29264 | closed | [
"usage"
] | 2025-11-23T11:45:54Z | 2025-11-24T13:03:50Z | 2 | josefdra |
vllm-project/vllm | 29,263 | [Feature]: Enable flash attention (and/or FlashMLA) for AMD GPUs | ### 🚀 The feature, motivation and pitch
In [this page from flash-attention](https://github.com/Dao-AILab/flash-attention?tab=readme-ov-file#amd-rocm-support), I checked that the upstream `flash-attention` currently has composable_kernel (for newer AMD GPUs) and WIP triton (for older RNDA GPUs, etc.) implementations. As well as [flash MLA](https://github.com/deepseek-ai/FlashMLA?tab=readme-ov-file#amd-instinct).
Is it possible to enable `vllm.vllm_flash_attn._vllm_fa2_C` and more modules for AMD GPUs?
### Alternatives
_No response_
### Additional context
_No response_
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29263 | closed | [
"feature request",
"rocm"
] | 2025-11-23T11:28:47Z | 2025-12-05T01:54:08Z | 4 | Inokinoki |
vllm-project/vllm | 29,245 | [Usage]: 启动 qwen3 vl 超级超级超级慢，sglang 启动很快，可能的原因是什么？ | ### Your current environment
连执行 python collect_env.py 都很慢，环境是直接 uv 安装的
```text
Collecting environment information...
==============================
System Info
==============================
OS : Ubuntu 24.04.2 LTS (x86_64)
GCC version : (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0
Clang version : Could not collect
CMake version : version 4.1.2
Libc version : glibc-2.39
==============================
PyTorch Info
==============================
PyTorch version : 2.9.0+cu128
Is debug build : False
CUDA used to build PyTorch : 12.8
ROCM used to build PyTorch : N/A
==============================
Python Environment
==============================
Python version : 3.12.3 (main, Jun 18 2025, 17:59:45) [GCC 13.3.0] (64-bit runtime)
Python platform : Linux-5.10.134-19.100.al8.x86_64-x86_64-with-glibc2.39
==============================
CUDA / GPU Info
==============================
Is CUDA available : True
CUDA runtime version : 12.9.86
CUDA_MODULE_LOADING set to :
GPU models and configuration :
GPU 0: NVIDIA L20Y
GPU 1: NVIDIA L20Y
GPU 2: NVIDIA L20Y
GPU 3: NVIDIA L20Y
GPU 4: NVIDIA L20Y
GPU 5: NVIDIA L20Y
GPU 6: NVIDIA L20Y
GPU 7: NVIDIA L20Y
Nvidia driver version : 570.148.08
cuDNN version : Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.9.10.2
/usr/lib/x86_64-linux-gnu/libcudnn_adv.so.9.10.2
/usr/lib/x86_64-linux-gnu/libcudnn_cnn.so.9.10.2
/usr/lib/x86_64-linux-gnu/libcudnn_engines_precompiled.so.9.10.2
/usr/lib/x86_64-linux-gnu/libcudnn_engines_runtime_compiled.so.9.10.2
/usr/lib/x86_64-linux-gnu/libcudnn_graph.so.9.10.2
/usr/lib/x86_64-linux-gnu/libcudnn_heuristic.so.9.10.2
/usr/lib/x86_64-linux-gnu/libcudnn_ops.so.9.10.2
HIP runtime version : N/A
MIOpen runtime version : N/A
Is XNNPACK available : True
==============================
CPU Info
==============================
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 46 bits physical, 57 bits virtual
Byte Order: Little Endian
CPU(s): 192
On-line CPU(s) list: 0-191
Vendor ID: GenuineIntel
Model name: Intel(R) Xeon(R) Platinum 8468V
CPU family: 6
Model: 143
Thread(s) per core: 2
Core(s) per socket: 48
Socket(s): 2
Stepping: 8
CPU(s) scaling MHz: 70%
CPU max MHz: 3800.0000
CPU min MHz: 800.0000
BogoMIPS: 4800.00
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 invpcid_single intel_ppin cdp_l2 ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req hfi avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq rdpid bus_lock_detect cldemote movdiri movdir64b enqcmd fsrm uintr md_clear serialize tsxldtrk pconfig arch_lbr amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
Virtualization: VT-x
L1d cache: 4.5 MiB (96 instances)
L1i cache: 3 MiB (96 instances)
L2 cache: 192 MiB (96 instances)
L3 cache: 195 MiB (2 instances)
NUMA node(s): 2
NUMA node0 CPU(s): 0-47,96-143
NUMA node1 CPU(s): 48-95,144-191
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; | https://github.com/vllm-project/vllm/issues/29245 | open | [
"usage"
] | 2025-11-22T20:41:27Z | 2025-12-11T11:23:54Z | 3 | hucorz |
huggingface/candle | 3,208 | `cudarc` dynamic loading support | Currently, `candle` uses `cudarc` with the `dynamic-linking` feature, which requires the executable to find the DLLs or SOs at startup. However, it would be more convenient if `candle` also supported the `dynamic-loading` feature from `cudarc` to load DLLs or SOs at runtime.
Is it possible for `candle` to support it? | https://github.com/huggingface/candle/issues/3208 | open | [] | 2025-11-22T18:18:25Z | 2025-11-25T09:00:27Z | 7 | mayocream |
huggingface/transformers | 42,331 | SAM3 does not support custom inference resolutions | ### System Info
Note: I am running the latest git version, sys Info should not be relevant to the issue
$ transformers env
Traceback (most recent call last):
File "/home/master-andreas/panopticon/test_env/bin/transformers", line 3, in <module>
from transformers.cli.transformers import main
File "/home/master-andreas/panopticon/test_env/lib/python3.12/site-packages/transformers/cli/transformers.py", line 23, in <module>
from transformers.cli.serve import Serve
File "/home/master-andreas/panopticon/test_env/lib/python3.12/site-packages/transformers/cli/serve.py", line 351, in <module>
class Serve:
File "/home/master-andreas/panopticon/test_env/lib/python3.12/site-packages/transformers/cli/serve.py", line 658, in Serve
) -> ChatCompletionChunk:
^^^^^^^^^^^^^^^^^^^
NameError: name 'ChatCompletionChunk' is not defined
### Who can help?
@yonigozlan
### Information
- [ ] The official example scripts
- [x] My own modified scripts
### Tasks
- [x] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [ ] My own task or dataset (give details below)
### Reproduction
```py
"""
Test script for SAM3 text prompting only.
This script demonstrates how to use SAM3 for text-based segmentation on images.
"""
import torch
from PIL import Image
import requests
from transformers import Sam3Processor, Sam3Model
import os
INFERENCE_RESOLUTION = (1008, 1008) # If run with anything else other than 1008 it fails
# INFERENCE_RESOLUTION = (1400, 1400)
def test_sam3_text_prompting():
"""Test SAM3 with text prompting on a sample image."""
# Set device
device = "cpu"
print(f"Using device: {device}")
# Load model and processor
print("Loading SAM3 model and processor...")
model = Sam3Model.from_pretrained("facebook/sam3").to(device)
processor = Sam3Processor.from_pretrained("facebook/sam3")
# Load a sample image
print("Loading sample image...")
image_url = "http://images.cocodataset.org/val2017/000000077595.jpg"
image = Image.open(requests.get(image_url, stream=True).raw).convert("RGB")
# Define text prompts to test
text_prompts = ["cat", "ear", "eye"]
for text_prompt in text_prompts:
print(f"\nTesting text prompt: '{text_prompt}'")
# Prepare inputs
inputs = processor(images=image, text=text_prompt, size=INFERENCE_RESOLUTION, return_tensors="pt").to(device)
# Run inference
with torch.no_grad():
outputs = model(**inputs)
# Post-process results
results = processor.post_process_instance_segmentation(
outputs,
threshold=0.5,
mask_threshold=0.5,
target_sizes=inputs.get("original_sizes").tolist()
)[0]
# Display results
num_objects = len(results['masks'])
print(f"Found {num_objects} objects matching '{text_prompt}'")
if num_objects > 0:
# Show scores for first few objects
scores = results['scores']
print(f"Confidence scores: {scores[:min(3, len(scores))].tolist()}")
# Show bounding boxes for first object
if 'boxes' in results and len(results['boxes']) > 0:
box = results['boxes'][0]
print(f"First object bounding box (xyxy): {box.tolist()}")
if __name__ == "__main__":
print("SAM3 Text Prompting Test Script")
print("=" * 40)
try:
test_sam3_text_prompting()
print("\n✓ All tests completed successfully!")
except Exception as e:
print(f"\n✗ Test failed with error: {e}")
raise
```
Output when INFERENCE_RESOLUTION=[1400, 1400]:
```sh
$ py test_sam3_text.py
SAM3 Text Prompting Test Script
========================================
Using device: cpu
Loading SAM3 model and processor...
Loading weights: 100%|█| 1468/1468 [00:00<00:00, 2709.52it/s, Materializing param=vision_encoder.neck.fpn
Loading sample image...
Testing text prompt: 'cat'
✗ Test failed with error: The size of tensor a (10000) must match the size of tensor b (5184) at non-singleton dimension 2
Traceback (most recent call last):
File "/home/master-andreas/panopticon/test_sam3_text.py", line 124, in <module>
test_sam3_text_prompting()
File "/home/master-andreas/panopticon/test_sam3_text.py", line 48, in test_sam3_text_prompting
outputs = model(**inputs)
^^^^^^^^^^^^^^^
File "/home/master-andreas/panopticon/test_env/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/master-andreas/panopticon/test_env/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/master-andreas/panopticon/test_env/lib/python3.12/site-packages/transformers/utils/generic.py", line 938, in wrapper
| https://github.com/huggingface/transformers/issues/42331 | closed | [
"bug"
] | 2025-11-21T22:17:08Z | 2025-12-10T22:46:39Z | 3 | Kallinteris-Andreas |
huggingface/lerobot | 2,500 | question about the gr00t policy | hi,
I see here https://huggingface.co/docs/lerobot/en/groot that gr00t is intergrated into lerobot.
is it in sync with the original repo: https://github.com/NVIDIA/Isaac-GR00T ?
I see in original repo that the dataset used to fine-tune, is a bit different from the original lerobot format, like libero dataset (https://huggingface.co/datasets/physical-intelligence/libero) used in pi model ,
therefore i wonder what dataset format should be used here in lerbot policy training ?
any example dataset that is passed to `--dataset.repo_id=$DATASET_ID` ?
is it a post-processed dataset ? | https://github.com/huggingface/lerobot/issues/2500 | open | [
"question",
"policies"
] | 2025-11-21T21:45:19Z | 2025-12-03T14:03:34Z | null | yanan1116 |
vllm-project/vllm | 29,192 | Tool Calling Parsers Fail to Populate tool_calls Array for Qwen2.5-Coder Models | # Tool Calling Parsers Fail to Populate `tool_calls` Array for Qwen2.5-Coder Models
## Environment
- **vLLM Version**: v0.11.2.dev115+g56669c1f2 (Blackwell build)
- **Model**: Qwen/Qwen2.5-Coder-14B-Instruct-AWQ
- **Quantization**: AWQ
- **Python Version**: 3.x (Docker container)
- **GPU**: NVIDIA GeForce RTX 5080 (16GB, Blackwell/sm_120)
- **Platform**: WSL2, Linux 6.6.87.2-microsoft-standard-WSL2
## Description
When using tool calling with Qwen2.5-Coder models, the model correctly generates tool calls in `<tools>` XML format, but both `qwen3_xml` and `qwen3_coder` parsers fail to extract these tool calls into the `tool_calls` array in the API response. The tool call information remains in the `content` field but the `tool_calls` array stays empty.
## Steps to Reproduce
1. Start vLLM with Qwen2.5-Coder and tool calling parser:
```bash
python -m vllm.entrypoints.openai.api_server \
--model Qwen/Qwen2.5-Coder-14B-Instruct-AWQ \
--quantization awq \
--enable-auto-tool-choice \
--tool-call-parser qwen3_xml # or qwen3_coder
```
2. Send a tool calling request:
```bash
curl -s http://localhost:8002/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "qwen2.5-coder-14b-awq",
"messages": [{"role": "user", "content": "What is the weather in San Francisco?"}],
"tools": [
{
"type": "function",
"function": {
"name": "get_weather",
"description": "Get the current weather for a location",
"parameters": {
"type": "object",
"properties": {
"location": {
"type": "string",
"description": "The city and state, e.g. San Francisco, CA"
}
},
"required": ["location"]
}
}
}
],
"tool_choice": "auto"
}'
```
## Actual Output
```json
{
"id": "chatcmpl-xxx",
"object": "chat.completion",
"model": "qwen2.5-coder-14b-awq",
"choices": [
{
"message": {
"role": "assistant",
"content": "<tools>\n{\n \"name\": \"get_weather\",\n \"arguments\": {\n \"location\": \"San Francisco, CA\"\n }\n}\n</tools>",
"tool_calls": []
}
}
]
}
```
## Expected Output
```json
{
"id": "chatcmpl-xxx",
"object": "chat.completion",
"model": "qwen2.5-coder-14b-awq",
"choices": [
{
"message": {
"role": "assistant",
"content": "",
"tool_calls": [
{
"type": "function",
"id": "call_0",
"function": {
"name": "get_weather",
"arguments": "{\"location\": \"San Francisco, CA\"}"
}
}
]
}
}
]
}
```
## Analysis
### Model Output (Correct)
The model correctly generates tool calls in the expected `<tools>` XML format:
```xml
<tools>
{
"name": "get_weather",
"arguments": {
"location": "San Francisco, CA"
}
}
</tools>
```
### Parser Behavior (Incorrect)
Both recommended parsers fail to extract tool calls:
- **hermes parser**: Expects `<tool_call>` tags, doesn't match `<tools>` tags
- **qwen3_xml parser**: Designed for `<tools>` tags but doesn't populate `tool_calls` array
- **qwen3_coder parser**: Also designed for Qwen but fails to populate array
### Root Cause
The parsers appear to load correctly (visible in logs as `'tool_call_parser': 'qwen3_xml'`) but the extraction logic fails to populate the OpenAI-compatible `tool_calls` array structure.
## Workaround
Manual extraction from the `content` field:
```python
import re
import json
def extract_tool_calls(response):
"""Extract tool calls from Qwen2.5-Coder <tools> tags"""
content = response['choices'][0]['message']['content']
pattern = r'<tools>\s*({.*?})\s*</tools>'
match = re.search(pattern, content, re.DOTALL)
if match:
tool_data = json.loads(match.group(1))
return [{
"type": "function",
"function": {
"name": tool_data["name"],
"arguments": json.dumps(tool_data["arguments"])
}
}]
return []
```
## Additional Context
### Multi-AI Consultation Results
Consulted with multiple AI models for parser recommendation:
- **Qwen3 Coder (480B)**: Recommended `qwen3_xml` parser
- **DeepSeek V3.1**: Ranked `qwen3_xml` (90% confidence), `qwen3_coder` (80% confidence)
- **Claude Sonnet 4.5**: Confirmed tag mismatch between Hermes and Qwen formats
All models agreed that the parser selection is correct, suggesting the issue is in the parser implementation rather than configuration.
### vLLM Configuration
```python
{
'tool_call_parser': 'qwen3_xml', # Confirmed in logs
'enable_auto_tool_choice': True,
'model': 'Qwen/Qwen2.5-Coder-14B-Instruct-AWQ',
'quantization': 'awq',
'max_model_len': 8192
}
```
## Impact
- **Severity**: High - Breaks OpenAI API compatibility for tool calling
- **Affected Models**: Likely all Qwen2.5-Coder variants
- | https://github.com/vllm-project/vllm/issues/29192 | open | [] | 2025-11-21T18:31:19Z | 2025-11-21T18:31:19Z | 0 | Platano78 |
vllm-project/vllm | 29,180 | [Bug]: Recorded `EngineCoreEventType.QUEUED` time is off | ### Your current environment
<details>
</details>
### 🐛 Describe the bug
When running benchmarking with the CLI:
- on one side the serving point `vllm serve ...`
- on the other side the benchmarking client : `vllm bench serve...`
(note that the two are running on the same machine, there is no networking delay)
I noticed that the `EngineCoreEventType.QUEUED` event recorder on the server side didn't match the time of posting the request. In my understanding these two should events should be approximately equivalent. These values aren't off by a few milliseconds, but here the mismatch can be pretty big, up to a few seconds.
I think the reason might be because adding [request to the scheduler](https://github.com/vllm-project/vllm/blob/fcb1d570bb8f95f5b7ded716a52fec902c535f0e/vllm/v1/core/sched/scheduler.py#L1166) cannot be done when the engine is running a decoding or a prefill, see the [`_process_input_queue` function](https://github.com/vllm-project/vllm/blob/fcb1d570bb8f95f5b7ded716a52fec902c535f0e/vllm/v1/engine/core.py#L801), where `add_request()` ultimately gets called. This can introduce delays before the queued event gets recorded, having "floating" requests that are not tracked in the logs.
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29180 | closed | [
"bug"
] | 2025-11-21T12:58:36Z | 2025-11-30T20:56:44Z | 4 | sducouedic |
vllm-project/vllm | 29,177 | [Usage]: Vllm + Intervl model local infra Image preprocessing / request adding becomes bottleneck even with more CPU cores — how to accelerate? | ### Your current environment
vllm 0.11.0
### How would you like to use vllm
### current phenomenon
When doing **batched image classification** (64 images per batch) with InternVL3_5-1B, the bottleneck is clearly in the **"Adding requests"** phase (image preprocessing).
Even after increasing CPU cores and setting `OMP_NUM_THREADS=16`, the preprocessing speed stays around **50 it/s**, while the actual generation phase is extremely fast (>1500 prompts/s).
```text
Adding requests: 100%|██████████| 64/64 [00:01<00:00, 52.67it/s] ← bottleneck
Processed prompts: 100%|█| 64/64 [00:00<00:00, 1515.23it/s, est. speed input: 812805.23 tok/s]
```
This means ~95% of the total latency is spent on CPU-side image preprocessing, （I have disabled dynamic resolution）
### Minimal Reproducible Example
```python
import os
from PIL import Image
from vllm import LLM, SamplingParams
from transformers import AutoTokenizer
model_path = "/data/code/haobang.geng/models/InternVL3_5-1B"
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
llm = LLM(
model=model_path,
dtype="bfloat16",
max_model_len=4096,
gpu_memory_utilization=0.95,
limit_mm_per_prompt={"image": 1},
trust_remote_code=True,
enforce_eager=False,
)
prompt = "<image>\nYou are an image classifier. Output only one word: safe or nsfw."
sampling_params = SamplingParams(temperature=0.0, max_tokens=8)
batch_inputs = []
for i in range(64):
img = Image.open(f"/path/to/images/{i}.jpg").convert("RGB")
batch_inputs.append({
"prompt": prompt,
"multi_modal_data": {"image": img},
})
outputs = llm.generate(batch_inputs, sampling_params=sampling_params, use_tqdm=True)
```
### Expected behavior
For pure-text batches, Adding requests is >2000 it/s such as qwen3vl.
Attempted solutions (all ineffective)
### my attempt to speed up
Increase CPU cores / set OMP_NUM_THREADS=16 → no speedup
mm_processor_kwargs={"max_dynamic_patch": 1, ...} → seems no speedup
Pre-resize images to 384×384 → helps a little (~55 it/s) but still far from ideal
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions.
Thank you for the great work on vLLM! Looking forward to a simple way to slove it | https://github.com/vllm-project/vllm/issues/29177 | open | [
"usage"
] | 2025-11-21T10:56:29Z | 2025-12-01T14:08:22Z | 3 | Passenger12138 |
huggingface/trl | 4,554 | Better packing of data with best-fit decrease strategy | Hello,
When using packing with the bfd strategy, it looks like too much truncation is done when the seq_length is smaller than the average length of the sequences we want to pack.
For example :
```python
from datasets import Dataset
from trl import pack_dataset
examples = {
"input_ids": [[1, 2, 3, 4], [5, 6], [7, 8, 9], [10]],
"attention_mask": [[1, 1, 1, 1], [1, 0], [1, 0, 0], [1]],
}
dataset = Dataset.from_dict(examples)
packed_dataset = pack_dataset(dataset, seq_length=3, strategy="bfd")
print(packed_dataset )
```
results in:
```python
{'input_ids': [[1, 2, 3], [7, 8, 9], [5, 6, 10]],
'attention_mask': [[1, 1, 1], [1, 0, 0], [1, 0, 1]],
'seq_lengths': [[3], [3], [2, 1]]}
```
So the token '4' is missing from the training tokens.
In a extreme case:
```python
examples_2 = {
"input_ids": [[0, 0], [1, 2, 3, 4], [5, 6, 7, 8, 9], [10]],
"attention_mask": [[1, 1], [1, 1, 1, 1], [1, 1, 1, 1, 1], [1]],
}
dataset_2 = Dataset.from_dict(examples_2)
print(pack_dataset(dataset_2, seq_length=1, strategy="bfd")[:])
```
results in:
```python
{'input_ids': [[0], [1], [5], [10]],
'attention_mask': [[1], [1], [1], [1]],
'seq_lengths': [[1], [1], [1], [1]]}
```
So here we are basically applying truncation to every sequence instead of having twelve sequences of one token.
If we put ourself in a more usefull setting, when I was finetunning on some very long sequences with a seq_lenfth of 4096, the majority of the tokens was discarded y the bfd packing. On my dataset, the bfd method kept only 0.2% of the total training tokens.
Is the behavior normal ?
I would find it useful to add an option to still have tokens that are deleted in other sequences, even if this is less than ideal. It would be a good compromise between the current versions of bfd and wrapped. | https://github.com/huggingface/trl/issues/4554 | closed | [
"✨ enhancement",
"❓ question"
] | 2025-11-21T07:53:55Z | 2025-12-16T20:37:02Z | 3 | ntnq4 |
vllm-project/vllm | 29,148 | [Usage]: Deployment of the embedding models | ### Your current environment
```text
==============================
System Info
==============================
OS : Ubuntu 22.04.5 LTS (x86_64)
GCC version : (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version : Could not collect
CMake version : version 3.22.1
Libc version : glibc-2.35
==============================
PyTorch Info
==============================
PyTorch version : 2.8.0+cu128
Is debug build : False
CUDA used to build PyTorch : 12.8
ROCM used to build PyTorch : N/A
==============================
Python Environment
==============================
Python version : 3.10.16 (main, Dec 11 2024, 16:24:50) [GCC 11.2.0] (64-bit runtime)
Python platform : Linux-5.15.0-161-generic-x86_64-with-glibc2.35
==============================
CUDA / GPU Info
==============================
Is CUDA available : True
CUDA runtime version : 12.8.61
CUDA_MODULE_LOADING set to : LAZY
GPU models and configuration :
GPU 0: NVIDIA GeForce RTX 5090
GPU 1: NVIDIA GeForce RTX 5090
GPU 2: NVIDIA GeForce RTX 5090
GPU 3: NVIDIA GeForce RTX 5090
GPU 4: NVIDIA GeForce RTX 5090
GPU 5: NVIDIA GeForce RTX 5090
GPU 6: NVIDIA GeForce RTX 5090
GPU 7: NVIDIA GeForce RTX 5090
Nvidia driver version : 570.172.08
cuDNN version : Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.9.7.0
/usr/lib/x86_64-linux-gnu/libcudnn_adv.so.9.7.0
/usr/lib/x86_64-linux-gnu/libcudnn_cnn.so.9.7.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_precompiled.so.9.7.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_runtime_compiled.so.9.7.0
/usr/lib/x86_64-linux-gnu/libcudnn_graph.so.9.7.0
/usr/lib/x86_64-linux-gnu/libcudnn_heuristic.so.9.7.0
/usr/lib/x86_64-linux-gnu/libcudnn_ops.so.9.7.0
HIP runtime version : N/A
MIOpen runtime version : N/A
Is XNNPACK available : True
```
### How would you like to use vllm
When deploying the embedding model, I found that the actual GPU memory usage included not only the model itself but also kv_cache. Is this a reasonable phenomenon? In version v0.9.0, the GPU memory usage was only for the model itself.
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29148 | closed | [
"usage"
] | 2025-11-21T03:57:59Z | 2025-11-21T06:17:18Z | 3 | Root970103 |
vllm-project/vllm | 29,139 | [Feature]: Optimize collectives in TP MoE case using torch.compile pass | ### 🚀 The feature, motivation and pitch
To avoid redundant work in MoE models in the TP case, sequence parallelism was added to the Deepseek model definition in #24134 and expanded to other models in #24982. However, to avoid performing surgery on the linear layer, the current approach performs more communication than necessary. With a torch.compile custom pass, we can rewrite the graph to remove the redundant computation.
### More details
Before the SP optimization, the ops in the model were:
```
- o_proj:[num_tokens, ...] -> [num_tokens, ...] (incomplete results)
- all_reduce:[num_tokens, ...] -> [num_tokens, ...]
- router:[num_tokens, ...] -> [num_tokens, ...]
- experts:[num_tokens, ...] -> [num_tokens, ...]
- ...
```
With sequence parallel enabled, this becomes:
```
- o_proj: [num_tokens, ...] -> [num_tokens, ...] (incomplete results)
- all_reduce: [num_tokens, ...] -> [num_tokens, ...]
- chunk: [num_tokens, ...] -> [num_tokens/tp, ...]
- router: [num_tokens/tp, ...] -> [num_tokens/tp, ...]
- experts: [num_tokens/tp, ...] -> [num_tokens/tp, ...]
- all_gather: [num_tokens/tp, ...] -> [num_tokens, ...]
```
Additionally, experts now properly do the dp+tp<->ep dispatch instead of just the original replicated dp<->ep dispatch.
Notice that the `all_reduce` does redundant communication as each TP rank only requires partial results. With a compile pass, we can convert the `all_reduce` -> `chunk` sequence into a `reduce_scatter`:
```
- o_proj: [num_tokens, ...] -> [num_tokens, ...] (incomplete results)
- reduce_scatter: [num_tokens, ...] -> [num_tokens/tp, ...]
- router: [num_tokens/tp, ...] -> [num_tokens/tp, ...]
- experts: [num_tokens/tp, ...] -> [num_tokens/tp, ...]
- all_gather: [num_tokens/tp, ...] -> [num_tokens, ...]
```
We should create a new `SequenceParallelismMoEPass`, controlled by a new `PassConfig.enable_sp_moe` flag (following the new naming convention in #27995) so that it can be turned on independently of regular SP. We will likely need to pad the number of tokens to a multiple of TP size, although like described in #29136, there are alternatives.
### Alternatives
Alternatively, the original optimization could be done as a compile pass as well, which would significantly clean up the MoE model definitions. However, that would mean that `VLLM_COMPILE` compilation mode would be required for this optimization and if compilation is disabled, the optimization would be disabled as well. Generally we accept lower performance in eager mode as compilation is on by default, but I know there was a reason this was done this way (don't remember why).
### Additional context
Original proposal comment: https://github.com/vllm-project/vllm/pull/24982#pullrequestreview-3259494618
cc @tlrmchlsmth @bnellnm @robertgshaw2-redhat @alexm-redhat @zou3519 @nvpohanh @youkaichao
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29139 | open | [
"help wanted",
"good first issue",
"performance",
"feature request",
"torch.compile"
] | 2025-11-21T01:36:06Z | 2025-12-07T15:39:48Z | 19 | ProExpertProg |
vllm-project/vllm | 29,097 | [Docs] Feedback for `/en/latest/` | ### 📚 The doc issue
no
### Suggest a potential alternative/fix
_No response_
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29097 | closed | [
"documentation"
] | 2025-11-20T14:53:44Z | 2025-11-21T07:51:57Z | 2 | ch950684-svg |
vllm-project/vllm | 29,089 | [Performance]: Can we use CUDA graph to accelerate the Qwen2_5omniAudioEncoder in Qwen2.5-Omni-3B? | ### Proposal to improve performance
<img width="3088" height="1264" alt="Image" src="https://github.com/user-attachments/assets/535d7854-b9db-4e40-8f85-1abe08b4d35e" />
The trace graph shows that Qwen2_5omniAudioEncoder has a large number of small kernel startups, indicating significant room for optimization.
Can we use CUDA graph to accelerate the Qwen2_5omniAudioEncoder in Qwen2.5-Omni-3B?
### Report of performance regression
_No response_
### Misc discussion on performance
_No response_
### Your current environment (if you think it is necessary)
```text
The output of `python collect_env.py`
```
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29089 | open | [
"performance"
] | 2025-11-20T12:13:58Z | 2025-11-20T12:13:58Z | 0 | xq25478 |
vllm-project/vllm | 29,078 | [Performance]: 多实例导致的cpu占用过高 | ### Your current environment
GPU: RTX4090
cuda version: cuda12.8
vllm version: 0.11.0
中文：我使用triton server的 vllm backend 启动了4个 minerU2.5 模型的实例，我的服务器上有2张卡，我每张卡启动了1个实例，我发现cpu负载有时候极高，几乎占满了我的服务器，我的服务器有96核，vllm backend使用的是AsyncLLMEngine，我观察到在单卡上启动一个实例时，我发送200张小尺寸的文字图做OCR时，fps可以达到最高，也就是每秒可以处理200张的图片，cpu负载在40-50%左右，为了进一步增加性能，我在两张卡上各启动了一个实例，但是我观察到此时cpu负载几乎达到99%，占用了极高的cpu，每个实例的fps只有120左右，性能几乎没有提升。
我做了大量的测试，我开始以为是triton server的问题，但经过排查，我认为问题可能出现在vllm推理时占用了很高的cpu，因为我不使用triton server，使用 `vllm serve`来模拟同样的情况，每个vllm实例推理时也占用掉了20-30%的cpu，如果这样，我的服务器即使有再多的GPU，也不能够提升模型的性能，我该如何调试？
english：
I launched 4 instances of the minerU2.5 model using the vllm backend of Triton Server. My server is equipped with 2 GPUs, with 1 instance running on each GPU. However, I noticed that the CPU load sometimes spikes to extremely high levels, nearly maxing out the server—which has 192 CPU cores. The vllm backend uses AsyncLLMEngine.
When running a single instance on one GPU and sending 200 small-sized text images for OCR, I achieved the highest FPS—processing up to 200 images per second—with the CPU load hovering around 40-50%. To further improve performance, I launched one instance on each of the two GPUs. But in this scenario, the CPU load reached nearly 99% (extremely high usage), and each instance only achieved around 120 FPS, with almost no performance gain.
I conducted numerous tests. Initially, I suspected the issue was with Triton Server, but after troubleshooting, I believe the problem lies in the high CPU usage during vllm inference. Even when not using Triton Server—simulating the same scenario with `vllm serve`—each vllm instance consumes 20-30% of the CPU. If this persists, adding more GPUs to the server will not improve model performance. How should I debug this?
### How would you like to use vllm
I want to run inference of a [[MinerU2.5-2509-1.2B]().](https://huggingface.co/opendatalab/MinerU2.5-2509-1.2B) I don't know how to integrate it with vllm.
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29078 | closed | [
"usage"
] | 2025-11-20T08:26:35Z | 2025-11-21T02:17:51Z | 4 | zjq1996518 |
huggingface/transformers | 42,291 | Can we disable IPython progress bar and use normal tqdm bar? | I like the normal tqdm bar much better, it is lighter, cleaner, simpler, and less stress on my eyes (no green color). I would love to have an option to use tqdm bar and not IPython bar. | https://github.com/huggingface/transformers/issues/42291 | closed | [] | 2025-11-20T01:26:11Z | 2025-12-28T08:02:45Z | 1 | weathon |
vllm-project/vllm | 29,023 | [Feature]: Disable logging `/metrics` | ### 🚀 The feature, motivation and pitch
- IGW hits `/metrics` continuously to understand the current load on the system
- This leads to an overload of logs
- We can disable this with `--disable-uvicorn-access-log`, but lose access to all access logs
We should have `--disable-uvicorn-metrics-access-log` to avoid logging * just * metrics. Per Gemini, we can do this with something like:
```python
# Define the routes for which access logs should be disabled
EXCLUDE_PATHS = ["/health", "/metrics"]
class EndpointFilter(logging.Filter):
def filter(self, record: logging.LogRecord) -> bool:
# Check if the log record contains arguments and if the path matches an excluded path
if record.args and len(record.args) >= 3:
path = record.args[2] # The path is typically the third argument in uvicorn access logs
if path in EXCLUDE_PATHS:
return False # Exclude this log record
return True # Include all other log records
```
Create a command line arg like `--disable-uvicorn-metrics-access-log`which selectively disables logging hits to `/metrics`
### Alternatives
_No response_
### Additional context
_No response_
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/29023 | open | [
"help wanted",
"good first issue",
"feature request"
] | 2025-11-19T18:25:48Z | 2025-11-19T21:57:34Z | 5 | robertgshaw2-redhat |
huggingface/sentence-transformers | 3,575 | How to override model's `max_seq_length`? | It seems that impossible to override model's max length from `sentence_bert_config.json`.
```python
from sentence_transformers import SentenceTransformer
m = SentenceTransformer("intfloat/e5-small", tokenizer_kwargs={"model_max_length":3})
print(m.tokenize(["hi hi hi hi hi hi hi hi hi hi hi hi hi"]))
# {'input_ids': tensor([[ 101, 7632, 7632, 7632, 7632, 7632, 7632, 7632, 7632, 7632, 7632, 7632,
# 7632, 7632, 102]]), 'token_type_ids': tensor([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]]), 'attention_mask': tensor([[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]])}
print(m.tokenize(["hi hi hi hi hi hi hi hi hi hi hi hi hi"], truncation=True))
# {'input_ids': tensor([[ 101, 7632, 7632, 7632, 7632, 7632, 7632, 7632, 7632, 7632, 7632, 7632,
# 7632, 7632, 102]]), 'token_type_ids': tensor([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]]), 'attention_mask': tensor([[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]])}
print(m[0].tokenizer(["hi hi hi hi hi hi hi hi hi hi hi hi hi"], truncation=True))
# {'input_ids': [[101, 7632, 102]], 'token_type_ids': [[0, 0, 0]], 'attention_mask': [[1, 1, 1]]}
m.max_seq_length = 3
print(m.tokenize(["hi hi hi hi hi hi hi hi hi hi hi hi hi"]))
# {'input_ids': tensor([[ 101, 7632, 102]]), 'token_type_ids': tensor([[0, 0, 0]]), 'attention_mask': tensor([[1, 1, 1]])}
```
This is happening because during load it load `max_seq_length` from `sentence_bert_config` and then in `Transformers` it will override `max_seq_length` only it wasn't set in `sentence_bert_config` https://github.com/huggingface/sentence-transformers/blob/ad28c0a982acc39c73abdf0019faca10f227ef28/sentence_transformers/models/Transformer.py#L101-L118 even if `model_max_length` is passed in `tokenizer_kwargs` and then `max_seq_length` will be used as `max_length` instead of passed in kwargs https://github.com/huggingface/sentence-transformers/blob/ad28c0a982acc39c73abdf0019faca10f227ef28/sentence_transformers/models/Transformer.py#L319-L327
Probably this can be fixed by
```diff
max_seq_length = min(max_seq_length, self.tokenizer.model_max_length)
```
Source https://github.com/embeddings-benchmark/mteb/pull/3587#discussion_r2542434603
I think this is cause of https://github.com/huggingface/sentence-transformers/issues/3187 | https://github.com/huggingface/sentence-transformers/issues/3575 | open | [] | 2025-11-19T16:42:27Z | 2025-11-20T13:47:13Z | null | Samoed |
huggingface/trl | 4,546 | Does TRL support PipelineRL for compute efficiency? | Hi 👋,
I'm trying to understand whether TRL currently supports (or plans to support) the PipelineRL approach described here:
- Paper: [https://arxiv.org/pdf/2509.19128v2](https://arxiv.org/pdf/2509.19128v2?utm_source=chatgpt.com)
- Overview: [https://arxiv.org/html/2509.19128](https://arxiv.org/html/2509.19128?utm_source=chatgpt.com)
PipelineRL introduces an actor–learner pipeline with in-flight weight updates, where actors keep generating while the learner updates weights concurrently. This reduces policy lag and improves GPU utilization for long-context RL runs.
Does TRL currently support this kind of pipelineRL workflow, or is there a recommended way to approximate it using the existing TRL trainers (GRPO + vLLM)?
If not, I'd love suggestions or best practices for building something similar on top of TRL.
Thanks! 🙏 | https://github.com/huggingface/trl/issues/4546 | open | [
"✨ enhancement",
"❓ question"
] | 2025-11-19T12:39:29Z | 2025-11-22T12:43:54Z | 3 | harisarang |
vllm-project/vllm | 28,996 | [Usage]: How to run a single data parallel deployment across multiple nodes without ray | ### Your current environment
2 Nodes, each node has 8 H20 GPUs.
### How would you like to use vllm
According to https://docs.vllm.ai/en/latest/serving/data_parallel_deployment/#internal-load-balancing
```shell
# node0
vllm serve Qwen3-Coder-480B-A35B-Instruct --trust-remote-code --max-num-seqs 64 --max-model-len 131072 --port $PORT0 --host :: --data-parallel-size 2 --data-parallel-size-local 1 --data-parallel-address $NODE0_IPV6 --data-parallel-rpc-port $PORT1
# node1
vllm serve Qwen3-Coder-480B-A35B-Instruct --trust-remote-code --max-num-seqs 64 --max-model-len 131072 --headless --data-parallel-size 2 --data-parallel-size-local 1 --data-parallel-start-rank 1 --data-parallel-address $NODE0_IPV6 --data-parallel-rpc-port $NODE0_PORT1
```
but all of them are hanging on waiting for init message from front-end.
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/28996 | closed | [
"usage"
] | 2025-11-19T06:47:22Z | 2025-11-27T06:17:22Z | 3 | crystalww |
vllm-project/vllm | 28,986 | [Feature]: Fused Kernel for GPT-OSS Router | ### 🚀 The feature, motivation and pitch
<img width="1257" height="250" alt="Image" src="https://github.com/user-attachments/assets/31eba061-522c-4521-b0a9-9f25bb36c3df" />
- Right now, we spend ~3.5% of the layer in the expert selection
- The operation is unfused
Write a fused kernel like we have for deepseek grouped_topk
### Alternatives
- torch compile
- triton
- cuda
### Additional context
_No response_
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/28986 | open | [
"help wanted",
"good first issue",
"feature request"
] | 2025-11-19T03:18:25Z | 2025-12-12T16:16:37Z | 7 | robertgshaw2-redhat |
huggingface/transformers.js | 1,458 | ONNX Backend Env variable | ### Question
Hi,
For some context, I'm building an application that uses some of the models on huggingface as an annotation tool that helps create annotations for training a specialised model.
As for the specialised model, I am able to export them to onnx, and I was able to run this model in the same application, but I have to manually install the same onnxruntime-web version to be able to do so. I looked into the docs [here](https://huggingface.co/docs/transformers.js/api/backends/onnx#module_backends/onnx.createInferenceSession), but I cannot access these functions through `env.backends.onnx`. I've tried `console.log(env.backends.onnx.isONNXProxy())` and got
```
Uncaught (in promise) TypeError: env.backends.onnx.isONNXProxy is not a function
```
Is there a way I can access the same inference session through this package?
---------------------------------
My `package.json`
```
{
"dependencies": {
"@huggingface/transformers": "3.7.5",
"onnxruntime-web": "1.22.0-dev.20250409-89f8206ba4"
},
}
``` | https://github.com/huggingface/transformers.js/issues/1458 | open | [
"question"
] | 2025-11-19T01:26:02Z | 2025-11-25T15:36:13Z | null | Heinrik-20 |
vllm-project/vllm | 28,956 | [Bug]: OOM when profiling multimodal model with multiple images | ### Your current environment
vLLM 0.11.0
### 🐛 Describe the bug
As per title.
The error log is as follows:
```
[multiproc_executor.py:671] Traceback (most recent call last):
[multiproc_executor.py:671] File "/root/miniconda3/lib/python3.11/site-packages/vllm/v1/executor/multiproc_executor.py", line 666, in worker_busy_loop
[multiproc_executor.py:671] output = func(*args, **kwargs)
[multiproc_executor.py:671] ^^^^^^^^^^^^^^^^^^^^^
[multiproc_executor.py:671] File "/root/miniconda3/lib/python3.11/site-packages/torch/utils/_contextlib.py", line 120, in decorate_context
[multiproc_executor.py:671] return func(*args, **kwargs)
[multiproc_executor.py:671] ^^^^^^^^^^^^^^^^^^^^^
[multiproc_executor.py:671] File "/root/miniconda3/lib/python3.11/site-packages/vllm/v1/worker/gpu_worker.py", line 263, in determine_available_memory
[multiproc_executor.py:671] self.model_runner.profile_run()
[multiproc_executor.py:671] File "/root/miniconda3/lib/python3.11/site-packages/vllm/v1/worker/gpu_model_runner.py", line 3379, in profile_run
[multiproc_executor.py:671] expanded = output.new_zeros(
[multiproc_executor.py:671] ^^^^^^^^^^^^^^^^^
[multiproc_executor.py:671] torch.OutOfMemoryError: CUDA out of memory. Tried to allocate 3.00 GiB. GPU 6 has a total capacity of 139.81 GiB of which 2.58 GiB is free. Including non-PyTorch memory, this process has 137.21 GiB memory in use. Of the allocated memory 134.77 GiB is allocated by PyTorch, and 255.64 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)
```
Looks like we only need **ONE** encoder cache with shape `(encoder_budget, encoder_output_shape[-1])` rather than `len(dummy_encoder_outputs)` ones.
https://github.com/vllm-project/vllm/blob/da8dadf68b5a2af849e7c5fd35ce9b8525d8d398/vllm/v1/worker/gpu_model_runner.py#L4128-L4144
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/28956 | closed | [
"bug"
] | 2025-11-18T17:36:55Z | 2025-11-25T12:38:37Z | 7 | imShZh |
huggingface/lerobot | 2,475 | Why there is difference between async inference and local inference in image resize? | I read code between `src/lerobot/async_inference/policy_server.py` and `src/lerobot/scripts/lerobot_record.py`. I found difference in these 2 code about inference which causes different image shape
1. `src/lerobot/scripts/lerobot_record.py` use this to deal with observation
And `prepare_observation_for_inference` is like this:
```python
def prepare_observation_for_inference(
observation: dict[str, np.ndarray],
device: torch.device,
task: str | None = None,
robot_type: str | None = None,
) -> RobotObservation:
for name in observation:
observation[name] = torch.from_numpy(observation[name])
if "image" in name:
observation[name] = observation[name].type(torch.float32) / 255
observation[name] = observation[name].permute(2, 0, 1).contiguous()
observation[name] = observation[name].unsqueeze(0)
observation[name] = observation[name].to(device)
observation["task"] = task if task else ""
observation["robot_type"] = robot_type if robot_type else ""
return observation
```
Here no **resize** operation in images i think.
2. in async_inference policy_server.py,it uses
But here function`prepare_raw_observation` makes sense on image shape
```python
def prepare_raw_observation(
robot_obs: RawObservation,
lerobot_features: dict[str, dict],
policy_image_features: dict[str, PolicyFeature],
) -> Observation:
"""Matches keys from the raw robot_obs dict to the keys expected by a given policy (passed as
policy_image_features)."""
# 1. {motor.pos1:value1, motor.pos2:value2, ..., laptop:np.ndarray} ->
# -> {observation.state:[value1,value2,...], observation.images.laptop:np.ndarray}
lerobot_obs = make_lerobot_observation(robot_obs, lerobot_features)
# 2. Greps all observation.images.<> keys
image_keys = list(filter(is_image_key, lerobot_obs))
# state's shape is expected as (B, state_dim)
state_dict = {OBS_STATE: extract_state_from_raw_observation(lerobot_obs)}
image_dict = {
image_k: extract_images_from_raw_observation(lerobot_obs, image_k) for image_k in image_keys
}
# Turns the image features to (C, H, W) with H, W matching the policy image features.
# This reduces the resolution of the images
image_dict = {
key: resize_robot_observation_image(torch.tensor(lerobot_obs[key]), policy_image_features[key].shape)
for key in image_keys
}
if "task" in robot_obs:
state_dict["task"] = robot_obs["task"]
return {**state_dict, **image_dict}
```
Here the shape of observation images is modified to policy config
```python
def resize_robot_observation_image(image: torch.tensor, resize_dims: tuple[int, int, int]) -> torch.tensor:
assert image.ndim == 3, f"Image must be (C, H, W)! Received {image.shape}"
# (H, W, C) -> (C, H, W) for resizing from robot obsevation resolution to policy image resolution
image = image.permute(2, 0, 1)
dims = (resize_dims[1], resize_dims[2])
# Add batch dimension for interpolate: (C, H, W) -> (1, C, H, W)
image_batched = image.unsqueeze(0)
# Interpolate and remove batch dimension: (1, C, H, W) -> (C, H, W)
resized = torch.nn.functional.interpolate(image_batched, size=dims, mode="bilinear", align_corners=False)
return resized.squeeze(0)
```
I found this when I can inference correctly locally by made weird action outputs from async inference. it must be caused by my didn't resize input image when training. -.-
version:deb9596bd3796c03ae3a5a6b81b63c1dba296256
| https://github.com/huggingface/lerobot/issues/2475 | open | [
"question"
] | 2025-11-18T14:32:17Z | 2025-11-24T02:23:13Z | null | milong26 |
vllm-project/vllm | 28,943 | [Usage]: what's the right way to run embedding model in vllm 0.11.0 | ### Your current environment
```text
The output of `python collect_env.py`
```
in vllm 0.8.7，I use following code to run local vllm，all is right：
```
self.engine_args = EngineArgs(
model=self.model_path,
dtype='half',
task="embed",
trust_remote_code=True,
limit_mm_per_prompt={"image": 1},
)
e = asdict(self.engine_args)
self.max_len = 100
self.llm = LLM(**e)
out = self.llm.embed(datas)
```
But in vllm 0.11.0 according to the document https://www.aidoczh.com/vllm/models/pooling_models.html，it use runner=='pooling' to run embedding task. What's the diffenence? Could the 'task' arg 'embed' still take effect?
### How would you like to use vllm
I want to run inference of a [specific model](put link here). I don't know how to integrate it with vllm.
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/28943 | open | [
"usage"
] | 2025-11-18T13:47:57Z | 2025-11-20T10:49:12Z | 3 | neverneverendup |
huggingface/trl | 4,541 | Is attn_implementation=sdpa not supported when using SFTTrainer with mllama? | When trying to use `sdpa` with mllama I get an error using the default collator. Upon writing my own collator it works.
When using `eager` implementation it gives cuda oom error. Is `sdpa` not supported? | https://github.com/huggingface/trl/issues/4541 | open | [] | 2025-11-18T11:57:01Z | 2025-11-18T11:57:01Z | 0 | osaidr |
vllm-project/vllm | 28,930 | [Usage]: How to build a qwen3vl embedding model with a custom mlp layer on the top use vllm? | ### Your current environment
```text
The output of `python collect_env.py`
```
Hi friends! I train a sft model built upon qwen3vl 2b model, we put a mlp layer on it to compress the embedding size of the backbone model. Now I want to use vllm 0.11.0 to serve it but I meet some confuse. Here is my custom class code
```
from argparse import Namespace
from dataclasses import asdict
from typing import Literal, NamedTuple, Optional, TypedDict, Union, get_args
import torch
import torch.nn as nn
from vllm.model_executor.models.qwen3_vl import Qwen3VLForConditionalGeneration
from vllm.v1.pool.metadata import PoolingMetadata
from vllm.v1.sample.metadata import SamplingMetadata
from vllm.config import VllmConfig
from vllm.multimodal import MULTIMODAL_REGISTRY
class CustomQwenVL3BPool(nn.Module):
def __init__(
self
):
super().__init__()
self.out = torch.nn.Sequential(
torch.nn.Linear(2048, 512),
torch.nn.SiLU(),
torch.nn.Linear(512, 128)
)
def get_prompt_lens(self,
hidden_states: Union[torch.Tensor, list[torch.Tensor]],
pooling_metadata: PoolingMetadata,
) -> torch.Tensor:
return pooling_metadata.prompt_lens
def forward(
self,
hidden_states: torch.Tensor,
pooling_metadata: PoolingMetadata,
) -> Union[list[torch.Tensor], torch.Tensor]:
# 1 提取lasttoken
prompt_lens = self.get_prompt_lens(hidden_states, pooling_metadata)
last_token_flat_indices = torch.cumsum(prompt_lens, dim=0) - 1
hidden_states = hidden_states[last_token_flat_indices]
# 2 mlp压缩维度
mlp_output = self.out(hidden_states)
# 3 正则化输出，需要check下vllm是否会再次norm
normalized_output = F.normalize(mlp_output, p=2, dim=-1)
return normalized_output
class CustomQwen3VLForConditionalGeneration(Qwen3VLForConditionalGeneration):
def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
super().__init__(vllm_config=vllm_config, prefix=prefix)
self._pooler = CustomQwenVL3BPool()
```
When I run above code using local mode of vllm , error log says **"[adapters.py:79] ST projector loading failed".Does anybody know why?** BTW，what's the best practice to make a custom embedding model with mlp in vllm 0.11.0
### How would you like to use vllm
I want to run inference of a [specific model](put link here). I don't know how to integrate it with vllm.
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/28930 | closed | [
"usage"
] | 2025-11-18T10:32:07Z | 2025-12-23T04:49:30Z | 10 | neverneverendup |
vllm-project/vllm | 28,929 | [Usage]: How | = | https://github.com/vllm-project/vllm/issues/28929 | closed | [
"usage"
] | 2025-11-18T10:26:17Z | 2025-11-18T10:30:53Z | 0 | neverneverendup |
huggingface/datasets | 7,869 | Why does dataset merge fail when tools have different parameters? | Hi, I have a question about SFT (Supervised Fine-tuning) for an agent model.
Suppose I want to fine-tune an agent model that may receive two different tools: tool1 and tool2. These tools have different parameters and types in their schema definitions.
When I try to merge datasets containing different tool definitions, I get the following error:
TypeError: Couldn't cast array of type
struct<refundFee: struct<description: string, type: string>, ... , servicerId: struct<description: string, type: string>>
to
{
'refundFee': {'description': Value(dtype='string'), 'type': Value(dtype='string')},
...
'templateId': {'description': Value(dtype='string'), 'type': Value(dtype='string')}
}
From my understanding, the merge fails because the tools column's nested structure is different across datasets — e.g., one struct contains an extra field servicerId while the other does not. This causes HuggingFace Datasets (and its underlying Apache Arrow schema) to reject the merge.
My question is: why is it designed this way?
Is this strict schema matching a hard requirement of the library?
Is there a recommended way to merge datasets with different tool schemas (different parameters and types)?
For an agent model supporting multiple tools, what's the best practice for preparing/merging training data without losing flexibility?
Any guidance or design rationale would be greatly appreciated. Thanks! | https://github.com/huggingface/datasets/issues/7869 | open | [] | 2025-11-18T08:33:04Z | 2025-11-30T03:52:07Z | 1 | hitszxs |
vllm-project/vllm | 28,903 | [Bug]: vllm inference on qwen3-vl when use_upstream_fa is False | ### Your current environment
pip show torch vllm flash-attn
Name: torch
Version: 2.8.0
---
Name: vllm
Version: 0.11.0
Name: flash_attn
Version: 2.8.3
### 🐛 Describe the bug
unit-test code as the follows,
when simple qwen3-0.6B can run; but qwen3-vl-4b not run
```python
#coding=utf-8
"""
写单元测试来验证FA和VLLM的可用性和兼容性
"""
import torch
from flash_attn import flash_attn_func
import unittest
import vllm
# from vllm.attention.backends import get_attn_backend
class TestFA_VLLM(unittest.TestCase):
def testFA(self,):
# 检查CUDA是否可用及设备
print(f"CUDA available: {torch.cuda.is_available()}")
print(f"Current device: {torch.cuda.current_device()}")
print(f"Device name: {torch.cuda.get_device_name()}")
# 尝试创建一个简单的张量并移动到GPU
try:
q = torch.randn(1, 1, 16, 64, dtype=torch.float16, device='cuda')
k = torch.randn(1, 1, 16, 64, dtype=torch.float16, device='cuda')
v = torch.randn(1, 1, 16, 64, dtype=torch.float16, device='cuda')
output = flash_attn_func(q, k, v)
print("FlashAttention test passed!")
except Exception as e:
print(f"FlashAttention test failed: {e}")
def oriTestVLLM(self,):
# 打印当前使用的attention后端
print("Available CUDA devices:", torch.cuda.device_count())
print("Current device:", torch.cuda.current_device())
print("Device name:", torch.cuda.get_device_name())
# 检查vLLM配置
print("vLLM version:", vllm.__version__)
# 尝试创建一个小模型来触发后端初始化
try:
from vllm import LLM
llm = LLM(model="Qwen/Qwen3-0.6B", max_model_len=256)
print("vLLM初始化成功!")
prompt = "这是一个测试提示。"
response = llm.generate(prompt)
print("rollout测试成功! 生成的文本:", response)
except Exception as e:
print(f"vLLM初始化失败: {e}")
def testVLLM(self,):
# 打印当前使用的attention后端
print("Available CUDA devices:", torch.cuda.device_count())
print("Current device:", torch.cuda.current_device())
print("Device name:", torch.cuda.get_device_name())
# 尝试创建一个小模型来触发后端初始化
try:
MODEL_PATH = "Qwen/Qwen3-VL-4B-Instruct"
from vllm import LLM
from vllm import LLM, SamplingParams
from vllm.assets.image import ImageAsset # vLLM 内置工具，帮你把路径 → PIL
from vllm.assets.video import VideoAsset # 如果以后想加视频同理
# 随便用一张图就行
image_path = ""
from PIL import Image
image = Image.open(image_path)
# 方式 B：URL
# image = ImageAsset("image", "https://xxx.jpg").pil_image
# Qwen3-VL 要求的对话模板
messages = [
{
"role": "user",
"content": [
{"type": "image", "image": image}, # 图像字段
{"type": "text", "text": "请描述这张图片。"}
]
}
]
# 用 transformers 的 apply_chat_template 把 messages → 模型输入
from transformers import AutoTokenizer
tok = AutoTokenizer.from_pretrained(MODEL_PATH)
prompt = tok.apply_chat_template(
messages,
tokenize=False,
add_generation_prompt=True
)
# ---------- ④ 生成 ----------
sampling_params = SamplingParams(
temperature=0.7,
max_tokens=512,
stop_token_ids=[tok.eos_token_id, tok.convert_tokens_to_ids("<|im_end|>")]
)
llm = LLM(model=MODEL_PATH, max_model_len=4096,
limit_mm_per_prompt={"image": 1, "video": 0}, # 每张 prompt 最多 1 张图
dtype="bfloat16", # A100/H100 可开；消费卡用 "float16"
gpu_memory_utilization=0.9,)
print("vLLM初始化成功!")
outputs = llm.generate(
{"prompt": prompt, "multi_modal_data": {"image": image}}, # 关键：把图也传进去
sampling_params=sampling_params
)
response = outputs[0].outputs[0].text
print("rollout测试成功! 生成的文本:", response)
except Exception as e:
print(f"vLLM初始化失败: {e}")
if __name__ == "__main__":
unittest.main()
```
error is :vllm/vllm_flash_attn/flash_attn_interface.py", line 233, in flash_attn_varlen_func [rank0]: out, softmax_lse = torch.ops._vllm_fa2_C.varlen_fwd( [rank0]: File "/usr/local/lib/python3.10/dist-packages/torch/_ops.py", line 1243, in __call__ [rank0]: return self._op(*args, **kwargs) [rank0]: torch.AcceleratorError: CUDA error: the provided PTX was compiled with an unsupported toolchain.
Then, I review the code in https://github.com/vllm-project/vllm/blob/main/vllm/model_executor/models/qwen3_vl.py#L375
it default set use_upstream_fa = False, when I change it to True, it works? the vllm version is 0.11.0
### Before submitting a new issue...
- | https://github.com/vllm-project/vllm/issues/28903 | closed | [
"bug"
] | 2025-11-18T03:54:11Z | 2025-11-18T08:18:09Z | 1 | hedes1992 |
huggingface/lerobot | 2,465 | loss:nan grdn:nan How to solve the gradient explosion problem in PI05 training? | When training Pi05 using Lerobot, has anyone encountered a situation where gradients explode immediately after training? Errors occur when the batch_size is set to 64 or 32. How can this be resolved?
Below are my training commands and error logs.
python src/lerobot/scripts/lerobot_train.py --dataset.repo_id=aa_merged280 --policy.type=pi05 \
--output_dir=./outputs/pi05_training2 --job_name=pi05_training2 \
--policy.pretrained_path=lerobot/pi05_base --policy.compile_model=true \
--policy.gradient_checkpointing=true --wandb.enable=true --policy.dtype=bfloat16 \
--steps=100000 --policy.device=cuda --batch_size=32 --policy.push_to_hub=false
INFO 2025-11-17 22:07:40 ot_train.py:351 step:200 smpl:6K ep:9 epch:0.03 loss:nan grdn:nan lr:2.5e-06 updt_s:4.478 data_s:0.038
WARNING 2025-11-17 22:07:40 db_utils.py:141 WandB logging of key "loss_per_dim" was ignored as its type "<class 'list'>" is not handled by this wrapper.
INFO 2025-11-17 22:22:38 ot_train.py:351 step:400 smpl:13K ep:18 epch:0.06 loss:nan grdn:nan lr:7.5e-06 updt_s:4.458 data_s:0.022
WARNING 2025-11-17 22:22:38 db_utils.py:141 WandB logging of key "loss_per_dim" was ignored as its type "<class 'list'>" is not handled by this wrapper.
INFO 2025-11-17 22:37:34 ot_train.py:351 step:600 smpl:19K ep:27 epch:0.10 loss:nan grdn:nan lr:1.3e-05 updt_s:4.456 data_s:0.022
WARNING 2025-11-17 22:37:34 db_utils.py:141 WandB logging of key "loss_per_dim" was ignored as its type "<class 'list'>" is not handled by this wrapper.
INFO 2025-11-17 22:52:31 ot_train.py:351 step:800 smpl:26K ep:36 epch:0.13 loss:nan grdn:nan lr:1.8e-05 updt_s:4.456 data_s:0.022
WARNING 2025-11-17 22:52:31 db_utils.py:141 WandB logging of key "loss_per_dim" was ignored as its type "<class 'list'>" is not handled by this wrapper.
INFO 2025-11-17 23:07:29 ot_train.py:351 step:1K smpl:32K ep:45 epch:0.16 loss:nan grdn:nan lr:2.3e-05 updt_s:4.459 data_s:0.022
| https://github.com/huggingface/lerobot/issues/2465 | open | [
"bug",
"policies",
"training"
] | 2025-11-18T03:46:28Z | 2025-12-03T16:13:56Z | null | Lilgeneric |
huggingface/lerobot | 2,464 | Questions about Pi0.5 Model Training Details and High Level Planning Implementation | Hello, while studying the Pi0.5 model, I have two questions regarding the model implementation that I would like to ask you:
1、The paper mentions that the model adopts two-stage pre-training and designs a comprehensive loss function. However, when checking the compute_loss part in the open-source code, it is found that currently only the action loss is calculated, and the loss related to the VLM (Vision-Language Model) in the pre-training stage is not reflected. I would like to confirm whether this part is implemented elsewhere in the code or if there are other design considerations?
2、The ablation experiments in the paper show that the jointly trained Pi0.5 performs excellently in explicit and implicit High Level planning, even better than GPT4 and manual upper-level planning. However, from the open-source model code, the implementation part related to the High Level planning step has not been found for the time being. I would like to know how this part of the function is reflected in the code?
Looking forward to your reply, thank you! | https://github.com/huggingface/lerobot/issues/2464 | open | [
"question",
"training"
] | 2025-11-18T01:27:59Z | 2025-11-20T10:45:34Z | null | Ginldaj |
vllm-project/vllm | 28,876 | [CI Failure]: should test_cumem.py use spawn or fork in cuda? | ### Name of failing test
tests/basic_correctness/test_cumem.py
### Basic information
- [ ] Flaky test
- [x] Can reproduce locally
- [ ] Caused by external libraries (e.g. bug in `transformers`)
### 🧪 Describe the failing test
The test only fails locally for me when I use vllm main branch and on the CI of my PR, error is caused by cuda tests using `fork` instead of `spawn` I think, in the CI, there is a line that's trying for force spawn: https://github.com/vllm-project/vllm/blob/f2b8e1c5510cf3621dc4b910f0eba5289d9fee88/.buildkite/test-pipeline.yaml#L99-L100, but looks like it's not effective. I looked at the function that decides to use fork or spawn: https://github.com/vllm-project/vllm/blob/f8b19c0ffd65f7f6f01a0da4a39b6890f5db40cb/tests/utils.py#L1027 and I don't think it looks like the flag `VLLM_WORKER_MULTIPROC_METHOD`. Although the issue doesn't repro in the main vllm CI. Wondering how do we fix this?
```
FAILED basic_correctness/test_cumem.py::test_python_error - RuntimeError: Cannot re-initialize CUDA in forked subprocess. To use CUDA with multiprocessing, you must use the 'spawn' start method
FAILED basic_correctness/test_cumem.py::test_basic_cumem - RuntimeError: Cannot re-initialize CUDA in forked subprocess. To use CUDA with multiprocessing, you must use the 'spawn' start method
FAILED basic_correctness/test_cumem.py::test_cumem_with_cudagraph - RuntimeError: Cannot re-initialize CUDA in forked subprocess. To use CUDA with multiprocessing, you must use the 'spawn' start method
FAILED basic_correctness/test_cumem.py::test_end_to_end[hmellor/tiny-random-LlamaForCausalLM] - RuntimeError: Cannot re-initialize CUDA in forked subprocess. To use CUDA with multiprocessing, you must use the 'spawn' start method
FAILED basic_correctness/test_cumem.py::test_end_to_end[facebook/opt-125m] - RuntimeError: Cannot re-initialize CUDA in forked subprocess. To use CUDA with multiprocessing, you must use the 'spawn' start method
FAILED basic_correctness/test_cumem.py::test_deep_sleep - RuntimeError: Cannot re-initialize CUDA in forked subprocess. To use CUDA with multiprocessing, you must use the 'spawn' start method
FAILED basic_correctness/test_cumem.py::test_deep_sleep_async - RuntimeError: Cannot re-initialize CUDA in forked subprocess. To use CUDA with multiprocessing, you must use the 'spawn' start method
```
### 📝 History of failing test
https://buildkite.com/vllm/ci/builds/39127/steps/canvas?jid=019a84f5-0fbf-46f3-859f-42c02a2d3de1
### CC List.
_No response_ | https://github.com/vllm-project/vllm/issues/28876 | open | [
"ci-failure"
] | 2025-11-17T18:58:08Z | 2025-11-17T20:59:14Z | 1 | jerryzh168 |
vllm-project/vllm | 28,868 | [Bug]: When compiling with ranges, we should pass the range information to Inductor | ### Your current environment
main
### 🐛 Describe the bug
Might be more of a feature request. Context is that https://github.com/vllm-project/vllm/pull/24248 adds a new compile ranges API, where a user can specify which ranges to compile on.
We should tell Inductor how to constrain the compilation on the symints of the compile ranges
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/28868 | open | [
"bug",
"torch.compile"
] | 2025-11-17T15:41:50Z | 2026-01-05T23:37:12Z | 1 | zou3519 |
vllm-project/vllm | 28,866 | [Usage]: When is going to be the next release? | Hi everyone,
Thank you for developing such a great tool!
I was wondering when the next release is scheduled. I’m interested in running Gemma3-text type architecture GGUF quantized models with VLLM. Are there any alternatives to do this with the latest release (v0.11.0)?
I also noticed that you merged this PR with the working solution on October 9:
https://github.com/vllm-project/vllm/pull/26189 | https://github.com/vllm-project/vllm/issues/28866 | open | [
"usage"
] | 2025-11-17T15:24:47Z | 2025-11-19T10:51:47Z | 1 | Invalid-coder |
huggingface/transformers | 42,241 | How to use padding with Mistral? | I'm trying to understand how to use Mistral with `batch_size` > 1. One aspect of this is setting `padding="longest"` in, e.g., `MistralCommonTokenizer.encode()`. But I'm getting `TypeError: 'set' object is not callable` when I try this. Example:
```python
import torch
from transformers import MistralForCausalLM, MistralCommonTokenizer
tokenizer = MistralCommonTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.3")
model = MistralForCausalLM.from_pretrained(
"mistralai/Mistral-7B-Instruct-v0.3",
dtype=torch.bfloat16,
attn_implementation="sdpa",
device_map="auto",
)
messages = [
"You are a pirate chatbot who always responds in pirate speak!",
"Who are you?",
]
model_inputs = tokenizer.encode(messages, return_tensors="pt", padding="longest").to(
model.device
)
```
Output:
```
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
Cell In[1], line 17
5 model = MistralForCausalLM.from_pretrained(
6 "mistralai/Mistral-7B-Instruct-v0.3",
7 dtype=torch.bfloat16,
8 attn_implementation="sdpa",
9 device_map="auto",
10 )
12 messages = [
13 "You are a pirate chatbot who always responds in pirate speak!",
14 "Who are you?",
15 ]
---> 17 model_inputs = tokenizer.encode(messages, return_tensors="pt", padding="longest").to(
18 model.device
19 )
21 generated_ids = model.generate(model_inputs, max_new_tokens=100, do_sample=True)
22 tokenizer.batch_decode(generated_ids)[0]
File ~/ad_hoc_analysis/src/asr_and_summarization/.venv/lib/python3.13/site-packages/transformers/tokenization_mistral_common.py:407, in MistralCommonTokenizer.encode(self, text, text_pair, add_special_tokens, padding, truncation, max_length, stride, pad_to_multiple_of, padding_side, return_tensors, verbose, **kwargs)
404 if text_pair:
405 raise ValueError("`MistralCommonTokenizer.encode` does not support `text_pair`.")
--> 407 padding_strategy, truncation_strategy, max_length, _ = self._get_padding_truncation_strategies(
408 padding=padding,
409 truncation=truncation,
410 max_length=max_length,
411 pad_to_multiple_of=pad_to_multiple_of,
412 verbose=verbose,
413 )
415 encoded_inputs = self._encode_plus(
416 text,
417 add_special_tokens=add_special_tokens,
(...) 429 verbose=verbose,
430 )
432 return encoded_inputs["input_ids"]
File ~/ad_hoc_analysis/src/asr_and_summarization/.venv/lib/python3.13/site-packages/transformers/tokenization_mistral_common.py:1034, in MistralCommonTokenizer._get_padding_truncation_strategies(self, padding, truncation, max_length, pad_to_multiple_of, verbose, **kwargs)
1031 max_length = self.model_max_length
1033 # Test if we have a padding token
-> 1034 if padding_strategy != PaddingStrategy.DO_NOT_PAD and (self.pad_token is None or self.pad_token_id < 0):
1035 raise ValueError(
1036 "Asking to pad but the tokenizer does not have a padding token. "
1037 "Please select a token to use as `pad_token` `(tokenizer.pad_token = tokenizer.eos_token e.g.)` "
1038 "or add a new pad token via `tokenizer.add_special_tokens({'pad_token': '[PAD]'})`."
1039 )
1041 # Check that we will truncate to a multiple of pad_to_multiple_of if both are provided
File ~/ad_hoc_analysis/src/asr_and_summarization/.venv/lib/python3.13/site-packages/transformers/tokenization_mistral_common.py:334, in MistralCommonTokenizer.pad_token(self)
329 @property
330 def pad_token(self) -> str:
331 """
332 String associated to the padding token in the vocabulary.
333 """
--> 334 return self.convert_ids_to_tokens(self.pad_token_id)
File ~/ad_hoc_analysis/src/asr_and_summarization/.venv/lib/python3.13/site-packages/transformers/tokenization_mistral_common.py:548, in MistralCommonTokenizer.convert_ids_to_tokens(self, ids, skip_special_tokens)
546 tokens: list[str] = []
547 for token_id in ids:
--> 548 if self._is_control_token(token_id) and skip_special_tokens:
549 continue
550 tokens.append(self.tokenizer.instruct_tokenizer.tokenizer.id_to_piece(token_id))
File ~/ad_hoc_analysis/src/asr_and_summarization/.venv/lib/python3.13/site-packages/transformers/tokenization_mistral_common.py:513, in MistralCommonTokenizer._is_control_token(self, token_id)
511 def _is_control_token(self, token_id: int) -> bool:
512 if self._tokenizer_type == MistralTokenizerType.spm:
--> 513 return token_id in self.tokenizer.instruct_tokenizer.tokenizer._control_tokens()
514 elif self._tokenizer_type == MistralTokenizerType.tekken:
515 return token_id < self.tokenizer.instruct_tokenizer.tokenizer.num_special_tokens
TypeError: 'set' object is not callable
```
Env:
```
- `transformers` version: 4.57.1 | https://github.com/huggingface/transformers/issues/42241 | closed | [] | 2025-11-17T12:54:21Z | 2025-11-19T06:11:44Z | null | TopCoder2K |
huggingface/chat-ui | 1,986 | HI i would like to use default_headers={ "X-HF-Bill-To": "org-name" } in my chatui local deployment how i can?? | Hi,
So i want to bill my Inference usage to my organization and like to pass default_headers={
"X-HF-Bill-To": "org-name"
} parameter how i can do that?? | https://github.com/huggingface/chat-ui/issues/1986 | open | [
"support"
] | 2025-11-17T08:33:41Z | 2025-11-17T08:33:41Z | null | aditya-oss-prog |
huggingface/diffusers | 12,672 | How to set pipe "requires_grad=true"？ | I have set the variable and the model "requires_grad=true" with the following:
` pipe.transformer.requires_grad = True
pipe.vae.requires_grad = True`
`prev_sample = prev_sample.detach().requires_grad_(True)`
but the "requires_grad" of result by the pipe is still not true:
`image_tar = pipe.vae.decode(prev_sample, return_dict=False)[0]`
"image_tar" still can not requires_grad, so how to set pipe "requires_grad=true"？(all the operation is during inference stage.) | https://github.com/huggingface/diffusers/issues/12672 | closed | [] | 2025-11-17T03:36:43Z | 2025-11-20T12:19:20Z | null | micklexqg |
huggingface/diffusers | 12,669 | Flux1-Dev inference with single file ComfyUI/SD-Forge Safetensors | Is it possible to run inference with diffusers using a single-file safetensors created for ComfyUI/SD-Forge?
It looks like FluxPipeline.from_single_file() might be intended for this purpose, but I'm getting the following errors:
```
import torch
from diffusers import FluxPipeline
pipe = FluxPipeline.from_single_file("./flux1-dev-fp8.safetensors", torch_dtype=torch.float8_e4m3fn, use_safetensors=True)
```
```
Traceback (most recent call last):
File "/home/user/flux/imgen.py", line 9, in <module>
pipe = FluxPipeline.from_single_file("./flux1-dev-fp8.safetensors", torch_dtype=torch.float8_e4m3fn, use_safetensors=True)
File "/home/user/.local/lib/python3.13/site-packages/huggingface_hub/utils/_validators.py", line 114, in _inner_fn
return fn(*args, **kwargs)
File "/home/user/.local/lib/python3.13/site-packages/diffusers/loaders/single_file.py", line 509, in from_single_file
loaded_sub_model = load_single_file_sub_model(
library_name=library_name,
...<11 lines>...
**kwargs,
)
File "/home/user/.local/lib/python3.13/site-packages/diffusers/loaders/single_file.py", line 127, in load_single_file_sub_model
loaded_sub_model = create_diffusers_t5_model_from_checkpoint(
class_obj,
...<4 lines>...
local_files_only=local_files_only,
)
File "/home/user/.local/lib/python3.13/site-packages/diffusers/loaders/single_file_utils.py", line 2156, in create_diffusers_t5_model_from_checkpoint
model.load_state_dict(diffusers_format_checkpoint)
~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/user/.local/lib/python3.13/site-packages/torch/nn/modules/module.py", line 2641, in load_state_dict
raise RuntimeError(
...<3 lines>...
)
RuntimeError: Error(s) in loading state_dict for T5EncoderModel:
Missing key(s) in state_dict: "encoder.embed_tokens.weight".
```
I checked the safetensors file and the T5 encoder is present. However, it is named differently, which confuses diffusers. | https://github.com/huggingface/diffusers/issues/12669 | open | [] | 2025-11-16T11:57:48Z | 2025-12-03T16:53:58Z | 12 | ddpasa |
huggingface/ai-deadlines | 41 | How to indicate ARR deadlines | Right now the yaml format assumes conferences with locations and dates, but ACL ARR has rolling deadlines not tied to a physical conference. We are largely operating around these deadlines. How can we incorporate these into this system? | https://github.com/huggingface/ai-deadlines/issues/41 | open | [] | 2025-11-15T00:26:33Z | 2025-11-15T00:26:33Z | null | morrisalp |
huggingface/diffusers | 12,662 | question on stable_audio_transformer.py | Execuse me, I am leaning the code of `class StableAudioDiTModel` , I do not know what is the argument ` global_states_input_dim` used to? It seems that it is a must component that should be packed before the hidden_states sequence. and its default dim seems larger then the transformer inner_dim. What is that componenet means? If it is used to take in additional conditions, that seems can be done in the encoder outside. and compared with the concatenate, I think it may be better to repeat condition embedding to the sequence length and concat on hidden_dim.
And what is the ` sample_size: int = 1024,` parameter used in the model creation? it seems not used during `forward` call
The func doc of `class StableAudioDiTModel:forward`, it said ``` encoder_attention_mask (`torch.Tensor` of shape `(batch_size, sequence_len)`, *optional*):```. why the shape of encoder_attention_mask is batch_size X sequence_len instead of batch_size X encoder_sequence_len to be identical with the shape of the input `encoder_hidden_states`
and why thee return value of this `forward` is the direct `(hidden_states,)` but not `(hidden_states * attention_mask, )`?
about the `class StableAudioDiTModel forward`, what is the shape of parameters `rotary_embedding` and `timestep`?
why the global_embedding is concated before the hidden_states? I think hidden_states is what we want to generated during DiT pipeline. while encoder_hidden_states is the condition signal, so global_embedding should be used to en-rich the encoder_hidden_states. and the action of concate the global_embedding before the input hidden_states sequence will change the input seq_length, according to[ [1]](https://github.com/Stability-AI/stable-audio-tools/blob/main/docs/conditioning.md#input-concatenation), the concatenation should be done in the feature_dim direction, is it?
It seems using normal LayerNorm layer instead of adaLN layer? | https://github.com/huggingface/diffusers/issues/12662 | open | [] | 2025-11-14T09:26:01Z | 2025-11-25T08:53:39Z | 1 | JohnHerry |
vllm-project/vllm | 28,717 | [Usage]: Errors running vLLM docker in a closed environment with gpt-oss-120b on RTX 6000 Pro | ### Your current environment
Can't get vLLM to start with the below configuration. Seems to have issues loading in the model .safetensors. Any ideas on what could be causing it?
vllm version: 0.11.1
CPU: Intel Xeon w7-2595X
GPU: NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition
Model: https://huggingface.co/openai/gpt-oss-120b/tree/main
Command:
docker run --rm --name vllm --gpus=all --runtime=nvidia -p 8000:8000 -e HF_HUB_OFFLINE=1 --ipc=host -v opt/models/cache/:/root/.cache/huggingface/hub vllm/vllm-openai:latest --model openai/gpt-oss-120b
Also tried:
docker run --rm --name vllm --gpus=all --runtime=nvidia -p 8000:8000 -e HF_HUB_OFFLINE=1 --ipc=host -v opt/models/cache/:/root/.cache/huggingface/hub vllm/vllm-openai:latest --model openai/gpt-oss-120b
with the same output.
Output:
INFO 11-12 06:23:18 [__init__.py:216] Automatically detected platform cuda.
[1;36m(APIServer pid=1)[0;0m INFO 11-12 06:23:21 [api_server.py:1839] vLLM API server version 0.11.0
[1;36m(APIServer pid=1)[0;0m INFO 11-12 06:23:21 [utils.py:233] non-default args: {'model': 'openai/gpt-oss-120b'}
[1;36m(APIServer pid=1)[0;0m INFO 11-12 06:23:21 [arg_utils.py:504] HF_HUB_OFFLINE is True, replace model_id [openai/gpt-oss-120b] to model_path [/root/.cache/huggingface/hub/models--openai--gpt-oss-120b/snapshots/b5c939de8f754692c1647ca79fbf85e8c1e70f8a]
[1;36m(APIServer pid=1)[0;0m `torch_dtype` is deprecated! Use `dtype` instead!
[1;36m(APIServer pid=1)[0;0m INFO 11-12 06:23:26 [model.py:547] Resolved architecture: GptOssForCausalLM
[1;36m(APIServer pid=1)[0;0m ERROR 11-12 06:23:26 [config.py:278] Error retrieving safetensors: Repo id must be in the form 'repo_name' or 'namespace/repo_name': '/root/.cache/huggingface/hub/models--openai--gpt-oss-120b/snapshots/b5c939de8f754692c1647ca79fbf85e8c1e70f8a'. Use `repo_type` argument if needed., retrying 1 of 2
[1;36m(APIServer pid=1)[0;0m ERROR 11-12 06:23:28 [config.py:276] Error retrieving safetensors: Repo id must be in the form 'repo_name' or 'namespace/repo_name': '/root/.cache/huggingface/hub/models--openai--gpt-oss-120b/snapshots/b5c939de8f754692c1647ca79fbf85e8c1e70f8a'. Use `repo_type` argument if needed.
[1;36m(APIServer pid=1)[0;0m INFO 11-12 06:23:28 [model.py:1730] Downcasting torch.float32 to torch.bfloat16.
[1;36m(APIServer pid=1)[0;0m INFO 11-12 06:23:28 [model.py:1510] Using max model len 131072
[1;36m(APIServer pid=1)[0;0m INFO 11-12 06:23:29 [scheduler.py:205] Chunked prefill is enabled with max_num_batched_tokens=8192.
[1;36m(APIServer pid=1)[0;0m INFO 11-12 06:23:29 [config.py:271] Overriding max cuda graph capture size to 992 for performance.
INFO 11-12 06:23:31 [__init__.py:216] Automatically detected platform cuda.
[1;36m(EngineCore_DP0 pid=308)[0;0m INFO 11-12 06:23:33 [core.py:644] Waiting for init message from front-end.
[1;36m(EngineCore_DP0 pid=308)[0;0m INFO 11-12 06:23:33 [core.py:77] Initializing a V1 LLM engine (v0.11.0) with config: model='/root/.cache/huggingface/hub/models--openai--gpt-oss-120b/snapshots/b5c939de8f754692c1647ca79fbf85e8c1e70f8a', speculative_config=None, tokenizer='/root/.cache/huggingface/hub/models--openai--gpt-oss-120b/snapshots/b5c939de8f754692c1647ca79fbf85e8c1e70f8a', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, tokenizer_revision=None, trust_remote_code=False, dtype=torch.bfloat16, max_seq_len=131072, download_dir=None, load_format=auto, tensor_parallel_size=1, pipeline_parallel_size=1, data_parallel_size=1, disable_custom_all_reduce=False, quantization=mxfp4, enforce_eager=False, kv_cache_dtype=auto, device_config=cuda, structured_outputs_config=StructuredOutputsConfig(backend='auto', disable_fallback=False, disable_any_whitespace=False, disable_additional_properties=False, reasoning_parser='openai_gptoss'), observability_config=ObservabilityConfig(show_hidden_metrics_for_version=None, otlp_traces_endpoint=None, collect_detailed_traces=None), seed=0, served_model_name=/root/.cache/huggingface/hub/models--openai--gpt-oss-120b/snapshots/b5c939de8f754692c1647ca79fbf85e8c1e70f8a, enable_prefix_caching=True, chunked_prefill_enabled=True, pooler_config=None, compilation_config={"level":3,"debug_dump_path":"","cache_dir":"","backend":"","custom_ops":[],"splitting_ops":["vllm.unified_attention","vllm.unified_attention_with_output","vllm.mamba_mixer2","vllm.mamba_mixer","vllm.short_conv","vllm.linear_attention","vllm.plamo2_mamba_mixer","vllm.gdn_attention","vllm.sparse_attn_indexer"],"use_inductor":true,"compile_sizes":[],"inductor_compile_config":{"enable_auto_functionalized_v2":false},"inductor_passes":{},"cudagraph_mode":[2,1],"use_cudagraph":true,"cudagraph_num_of_warmups":1,"cudagraph_capture_sizes":[992,976,960,944,928,912,896,880,864,848,832,816,800,784,768,752,736,720,704,688,672,656,640,624,608,592,576,560,544,528,512,496,480,464,448,432,416,400,384,368,352,336,320,304,288,272,256,248,240,232,224,216,208,200,192,184,176,168,160,152,144,136,128,120,112,104,96,88,80,72,64,56,48 | https://github.com/vllm-project/vllm/issues/28717 | open | [
"usage"
] | 2025-11-14T08:49:48Z | 2025-11-20T15:45:21Z | 3 | antonkarlsson1 |
huggingface/trl | 4,525 | How to modify the advantage computation in GRPOTrainer | I’m looking to customize the advantage computation used in the DAPO algorithm. Do I need to subclass the full GRPOTrainer to do this, or is it sufficient to overwrite the logic in _generate_and_score_completions, since that method appears to handle the advantage calculation? | https://github.com/huggingface/trl/issues/4525 | open | [
"❓ question",
"🏋 GRPO"
] | 2025-11-14T03:48:17Z | 2025-11-14T11:37:18Z | null | Tuziking |
huggingface/transformers | 42,200 | Request of rewriting implementation of prediction_step in trainer.py | ### System Info
Any system. Because it's a problem coming from source code.
### Who can help?
@SunMarc
### Information
- [ ] The official example scripts
- [x] My own modified scripts
### Tasks
- [ ] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [x] My own task or dataset (give details below)
### Reproduction
Hi, i am talking about an issue that was reported 5 years ago but still exists in 2025, specifically, 13th Nov, 2025.
I quote one of the issues that was discussed before, ignored by sgugger. Please find the link below
https://discuss.huggingface.co/t/cuda-out-of-memory-when-using-trainer-with-compute-metrics/2941
When i was about to fine tune a LLM today, i ran into the same issue but i got saved by one folk's solution provided in this discussion.
How to reproduce (you should have a GPU, no quantization, just full fine tuning):
1. Find a random decoder-only text2text LLM, let's say Qwen3 0.6B.
2. Prepare a train dataset (>0 rows) and eval dataset (>850 rows).
3. Set eval_on_start = True, either TrainingArguments or SFTConfig could work.
4. Implement your own compute_metrics BUT DON'T implement preprocess_logits_for_metrics.
5. start training (don't need deepspeed or accelerate, just trainer.train())
What would happen?
First it would go through the evaluation dataset because i set eval_on_start=True, the model would go really fast originally but then it would go extremely slow. Finally, you would get an error that says numpy is trying to allocate a ridiculously big array to memory.
<img width="1567" height="986" alt="Image" src="https://github.com/user-attachments/assets/e1885324-fb09-48b6-8bfd-d36306c2a156" />
One of the folk who seems to be inspired by example code provided the implementation of preprocess_logits_for_metrics, which solved problem i encountered perfectly. The evaluation run is done within 2 mins.
Why it would happen?
I briefly go over the source code of evaluation_loop and i located prediction_step.
prediction_step says it would return a tuple of three optional torch.Tensor (loss, logits, label).
<img width="719" height="68" alt="Image" src="https://github.com/user-attachments/assets/537032b1-9371-4852-bed8-8f31cd6a0437" />
But most of the time, the returned logits is a tuple.
Why?
if you look at the the function that processes logits before logits is returned:
<img width="535" height="140" alt="Image" src="https://github.com/user-attachments/assets/d6f7f3b1-6c2a-4298-b4c2-f0ab85fa88cf" />
This function would receive all kinds of "tensors". The type of "tensors" could be list, tuple, Mapping or torch.Tensor.
Does it change the variable, called "tensors", from other data types to torch.Tensor?
No.
type(tensors)(........) would preserve the original type of tensors. It means if the variable "tensors" (i hate this variable name because it is misleading and confusing) is a tuple, after this function, it's still a tuple!!!!!
It's a recursive function btw. I would love doing recursion in programming competition, but not in huggingface codebase!!! It also implies a fact that the input of nested_detach could be complexly nested, like ([],())
So this function doesn't guarantee the logits is a torch.Tensor.
Nor does the implementation of prediction_step before nested_detach was called in prediction_step
<img width="702" height="759" alt="Image" src="https://github.com/user-attachments/assets/451982b4-648b-4876-a2b7-c9d748899fd1" />
So, the logits is not always a torch.Tensor, which is contradictory to what the type hint says, what did developers do?
They developed preprocess_logits_for_metrics.
So that user could fix it ON THEIR OWN IMPLEMENTATION.
(preprocess_logits_for_metrics is called within evaluation_loop to clean the mess, specifically, logits, returned by prediction_step())
<img width="803" height="772" alt="Image" src="https://github.com/user-attachments/assets/c7494018-e282-4577-b824-3db9c9e57609" />
It's such a lazy fix. Why a regular user is expected to implement their own preprocess_logits_f
or_metrics, to deal with a poorly-designed prediction_step?
It has been 5 years since the person who reported it.........
If a user-defined compute_metrics is not provided to Trainer or SFTTrainer, the prediction_step would return (loss, none, none), which skips the whole problem and this is why users said the issue of "slow evaluation" is gone when they don't provide compute_metrics.
I would like to make a Pull Request to fix it but i don't have enough time and energy to do this massive amount of work.
A temporary fix is to let users know when they need to make their own compute_metrics, they also have to implement preprocess_logits_for_metrics. Different models would have different styles of implementations but for text2text decoder only LLM.
<img width="687" height="78" alt="Image" src="https://github.com/user-attachments/assets/125fffe3-d8cc-44c7-9a96-35a11500d975" />
(Another thing is that the variable called | https://github.com/huggingface/transformers/issues/42200 | open | [
"Good Second Issue",
"bug"
] | 2025-11-14T00:13:40Z | 2025-12-18T14:29:32Z | 3 | Yacklin |
huggingface/transformers | 42,197 | Attempt to access socket despite HF_HUB_OFFLINE = 1 if cache warmed outside current process | ### System Info
- `transformers` version: 4.57.1
- Platform: Linux-6.6.84.1-microsoft-standard-WSL2-x86_64-with-glibc2.35
- Python version: 3.13.0
- Huggingface_hub version: 0.36.0
- Safetensors version: 0.6.2
- Accelerate version: not installed
- Accelerate config: not found
- DeepSpeed version: not installed
- PyTorch version (accelerator?): 2.9.1+cpu (NA)
- Tensorflow version (GPU?): not installed (NA)
- Flax version (CPU?/GPU?/TPU?): not installed (NA)
- Jax version: not installed
- JaxLib version: not installed
- Using distributed or parallel set-up in script?: No
### Who can help?
@ydshieh I have created a reproducible example of the issue I mentioned in https://github.com/huggingface/transformers/issues/41311#issuecomment-3508674325.
### Information
- [ ] The official example scripts
- [x] My own modified scripts
### Tasks
- [ ] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [x] My own task or dataset (give details below)
### Reproduction
Reproducible example: https://github.com/fr1ll/HF_HUB_OFFLINE
Warming the cache in a subprocess, then disabling sockets, then loading the same model should work.
However, it fails with an attempt to access a socket and then "Can't load" errors.
The script named `subprocess-warm_then_offline-load.py` reproduces this error.
Interestingly, warming the cache in process, then disabling sockets, then loading the same model works.
This is reproduced in `inprocess-warm_then_offline-load.py` in the repo above.
### Expected behavior
When a model has already been loaded into the cache ("warm cache"), if `HF_OFFLINE_MODE` = `"1"`, a Transformers pipeline should be able to load the model without accessing any network sockets. | https://github.com/huggingface/transformers/issues/42197 | closed | [
"Good Second Issue",
"bug"
] | 2025-11-13T21:38:29Z | 2025-11-24T09:33:54Z | 6 | fr1ll |
vllm-project/vllm | 28,646 | [Feature][P2]: Implement CI Build Time and Size Guards | ### 🚀 The feature, motivation and pitch
### Description
Once we optimize the Docker build, we need to prevent regressions. Create CI checks that fail if build time exceeds thresholds or if image size grows beyond acceptable limits. Also set up monitoring dashboards.
### What You'll Do
1. Create Python scripts to check image metrics:
- `check-image-size.py` (extend existing wheel size checker)
- `check-build-time.py`
- `check-image-layers.py`
2. Add these checks to CI pipeline after image build
3. Set appropriate thresholds (configurable)
4. Create Buildkite annotations for warnings
5. Set up CloudWatch dashboard for metrics (optional)
### Deliverables
- [ ] Python scripts for checking metrics
- [ ] Integration into test-template-ci.j2
- [ ] Configurable thresholds via environment variables
- [ ] Documentation on how to adjust thresholds
- [ ] CloudWatch dashboard (optional)
### Alternatives
_No response_
### Additional context
_No response_
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/28646 | open | [
"feature request",
"ci/build"
] | 2025-11-13T12:50:34Z | 2025-11-13T18:55:29Z | 0 | rzabarazesh |
huggingface/diffusers | 12,650 | Question about the `# Copied from` system | Hi team! 👋
While working on improving docstrings and type hints across scheduler files (issue #9567), I've noticed the `# Copied from` pattern used extensively throughout the codebase.
Examples:
- Functions like `betas_for_alpha_bar` are duplicated across multiple schedulers
- Output classes like `DDPMSchedulerOutput` are copied with name replacements (e.g., DDPM->EulerDiscrete)
My question: What's the rationale behind this duplication system instead of:
1. Using a shared utils.py or common.py file for common functions
2. Using class inheritance for similar Output classes
I understand there might be good architectural reasons (module independence, API stability, avoiding circular dependencies, etc.), but this isn't documented anywhere that I could find.
Suggested action: Regardless of the answer, I think we should either:
- Option A: Refactor to use inheritance/shared utilities (if the current system is legacy)
- Option B: Document this design decision in:
- A CONTRIBUTING.md or architecture doc
- Comments in the utils/check_copies.py script itself
- Another README in the diffusers directory
This would help future contributors (like me! 😅) understand why this pattern exists and how to work with it properly when improving documentation. What do you think?
Thanks for maintaining such a great library! 🚀 | https://github.com/huggingface/diffusers/issues/12650 | open | [] | 2025-11-13T11:53:22Z | 2025-12-21T22:44:03Z | 3 | delmalih |
huggingface/transformers | 42,179 | Add TileLang Kernel Support | ### Feature request
I would like to propose adding support for TileLang kernel in the transformers library. TileLang is a modular approach for writing attention kernels that could provide flexibility and performance benefits.
github link: https://github.com/tile-ai/tilelang
- Add TileLang as an optional attention backend
- Provide configuration options similar to existing attention mechanisms
- Ensure compatibility with existing model architectures
- Add proper multi-GPU support and synchronization
### Motivation
- Enhanced Modularity
TileLang offers a more modular approach to writing attention kernels, making it easier for researchers and developers to modify and optimize the implementation for specific use cases.
- Performance Comparison
Integrating TileLang would allow users to benchmark its performance directly against existing attention implementations, such as Flex Attention and Flash Attention. This would foster a better understanding of how different kernels can impact model performance and efficiency.
- Community Engagement
Supporting TileLang in the Transformers library would attract a broader community of developers interested in optimizing transformer models, thus enhancing collaboration and innovation.
- Flexibility
TileLang's architecture is designed for ease of modification, allowing users to experiment with and refine attention mechanisms more effectively.
### Your contribution
I've experimented with TileLang kernel on transformers models and found it works well in single-GPU scenarios. However, when enabling multi-GPU inference using `device_map='auto'`, I encounter NaN tensors. This may be related to tensor synchronization issues in distributed settings.
I'm willing to help with testing and potentially contributing to the implementation once the multi-GPU synchronization issue is understood and resolved.
I also have 3 questions:
1. Is there any existing plan or roadmap for TileLang integration?
2. Are there specific guidelines for adding new attention backends?
3. What would be the recommended approach for handling multi-GPU synchronization in custom kernels?
| https://github.com/huggingface/transformers/issues/42179 | open | [
"Feature request"
] | 2025-11-13T11:38:33Z | 2025-11-13T11:38:33Z | 0 | crownz248 |
huggingface/tokenizers | 1,885 | Feature request: Characters delimiter argument | I wish to develop a k-mer-character-based BPE tokenizer using your beautiful Rust package, for genomic applications. Unfortunately, it doesn't seem to support defining a characters delimiter. As I see it, it is a pretty straightforward change, instead of iterating a word by character, first split it by the delimiter and then iterate. Also, when merges are computed, in the string representation the character delimiter should also be considered. In that way, a multi-character word splitting could have been made feasible. Right now I am using a modified Python version of the BPE tokenizer made by the genius [Yikai-Liao](https://github.com/Yikai-Liao/efficient_bpe/blob/main/ebpe_v2.py), however it would be nice to see that happening in Rust as well, and natively supported by huggingface. Unfortunately, I am still novice in working with Rust, otherwise I would make a pull request with the suggested changes. Is it something that can be worked out in the future? Or is there a way to do this with the current implementation? Thank you! | https://github.com/huggingface/tokenizers/issues/1885 | open | [] | 2025-11-13T10:40:29Z | 2025-11-28T07:51:07Z | 1 | VasLem |
vllm-project/vllm | 28,629 | [Usage]: TPOT per request information was not collected by vllm bench serve | ### Your current environment
```text
The output of `python collect_env.py`
Collecting environment information...
==============================
System Info
==============================
OS : Ubuntu 24.04.2 LTS (x86_64)
GCC version : (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0
Clang version : Could not collect
CMake version : version 4.1.0
Libc version : glibc-2.39
==============================
PyTorch Info
==============================
PyTorch version : 2.8.0+xpu
Is debug build : False
CUDA used to build PyTorch : None
ROCM used to build PyTorch : N/A
==============================
Python Environment
==============================
Python version : 3.12.3 (main, Aug 14 2025, 17:47:21) [GCC 13.3.0] (64-bit runtime)
Python platform : Linux-6.14.0-1006-intel-x86_64-with-glibc2.39
==============================
CUDA / GPU Info
==============================
Is CUDA available : False
CUDA runtime version : No CUDA
CUDA_MODULE_LOADING set to : N/A
GPU models and configuration : No CUDA
Nvidia driver version : No CUDA
cuDNN version : No CUDA
HIP runtime version : N/A
MIOpen runtime version : N/A
Is XNNPACK available : True
==============================
CPU Info
==============================
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 46 bits physical, 57 bits virtual
Byte Order: Little Endian
CPU(s): 32
On-line CPU(s) list: 0-31
Vendor ID: GenuineIntel
BIOS Vendor ID: Intel(R) Corporation
Model name: Intel(R) Xeon(R) w5-3435X
BIOS Model name: Intel(R) Xeon(R) w5-3435X CPU @ 3.1GHz
BIOS CPU family: 179
CPU family: 6
Model: 143
Thread(s) per core: 2
Core(s) per socket: 16
Socket(s): 1
Stepping: 8
CPU(s) scaling MHz: 45%
CPU max MHz: 4700.0000
CPU min MHz: 800.0000
BogoMIPS: 6192.00
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 intel_ppin cdp_l2 ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect user_shstk avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req vnmi avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq la57 rdpid bus_lock_detect cldemote movdiri movdir64b enqcmd fsrm md_clear serialize tsxldtrk pconfig arch_lbr ibt amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
Virtualization: VT-x
L1d cache: 768 KiB (16 instances)
L1i cache: 512 KiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 45 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-31
Vulnerability Gather data sampling: Not affected
Vulnerability Ghostwrite: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and | https://github.com/vllm-project/vllm/issues/28629 | open | [
"usage"
] | 2025-11-13T09:20:19Z | 2025-11-13T09:20:19Z | 0 | jlwang1996 |
vllm-project/vllm | 28,626 | [Bug]:Qwen3-VL-32B-AWQ model memory usage: 8k context limit with 40GB VRAM? | ### Your current environment
<details>
<summary>The output of <code>python collect_env.py</code></summary>
```text
Your output of `python collect_env.py` here
```
</details>
### 🐛 Describe the bug
Running models on the latest stable vLLM release: https://huggingface.co/QuantTrio/Qwen3-VL-32B-Instruct-AWQ
The model size is 20GB, and my GPU has 40GB VRAM total.
Using parameter: --gpu-memory-utilization 0.9
Why am I only getting around 8k max context length? Do VL models really hog that much VRAM?
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/28626 | open | [
"bug"
] | 2025-11-13T08:00:20Z | 2025-11-17T07:08:47Z | 3 | maxin9966 |
vllm-project/vllm | 28,622 | [Bug]: Can we able to benchmark Quantized MOE models Either W8A8 or W8A16 ? | ### Your current environment
<details>
<summary>The output of <code>python collect_env.py</code></summary>
```text
Collecting environment information...
==============================
System Info
==============================
OS : Ubuntu 24.04.2 LTS (x86_64)
GCC version : (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0
Clang version : Could not collect
CMake version : version 3.22.1
Libc version : glibc-2.39
==============================
PyTorch Info
==============================
PyTorch version : 2.8.0+cu128
Is debug build : False
CUDA used to build PyTorch : 12.8
ROCM used to build PyTorch : N/A
==============================
Python Environment
==============================
Python version : 3.10.18 (main, Jun 4 2025, 08:56:00) [GCC 13.3.0] (64-bit runtime)
Python platform : Linux-6.14.0-33-generic-x86_64-with-glibc2.39
==============================
CUDA / GPU Info
==============================
Is CUDA available : True
CUDA runtime version : 12.0.140
CUDA_MODULE_LOADING set to : LAZY
GPU models and configuration : GPU 0: NVIDIA RTX 6000 Ada Generation
Nvidia driver version : 575.57.08
cuDNN version : Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.9.11.0
/usr/lib/x86_64-linux-gnu/libcudnn_adv.so.9.11.0
/usr/lib/x86_64-linux-gnu/libcudnn_cnn.so.9.11.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_precompiled.so.9.11.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_runtime_compiled.so.9.11.0
/usr/lib/x86_64-linux-gnu/libcudnn_graph.so.9.11.0
/usr/lib/x86_64-linux-gnu/libcudnn_heuristic.so.9.11.0
/usr/lib/x86_64-linux-gnu/libcudnn_ops.so.9.11.0
HIP runtime version : N/A
MIOpen runtime version : N/A
Is XNNPACK available : True
==============================
CPU Info
==============================
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 52 bits physical, 57 bits virtual
Byte Order: Little Endian
CPU(s): 52
On-line CPU(s) list: 0-51
Vendor ID: GenuineIntel
Model name: Intel(R) Xeon(R) w7-2595X
CPU family: 6
Model: 143
Thread(s) per core: 2
Core(s) per socket: 26
Socket(s): 1
Stepping: 8
CPU(s) scaling MHz: 21%
CPU max MHz: 4800.0000
CPU min MHz: 800.0000
BogoMIPS: 5616.00
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 intel_ppin cdp_l2 ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect user_shstk avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req vnmi avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq la57 rdpid bus_lock_detect cldemote movdiri movdir64b enqcmd fsrm md_clear serialize tsxldtrk pconfig arch_lbr ibt amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
Virtualization: VT-x
L1d cache: 1.2 MiB (26 instances)
L1i cache: 832 KiB (26 instances)
L2 cache: 52 MiB (26 instances)
L3 cache: 48.8 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-51
Vulnerability Gather data sampling: Not affected
Vulnerability Ghostwrite: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not af | https://github.com/vllm-project/vllm/issues/28622 | open | [
"bug"
] | 2025-11-13T07:26:56Z | 2025-11-13T07:27:06Z | 0 | logesh13 |
vllm-project/vllm | 28,610 | [Usage]: Does 0.11.0 suport tree attenton with eagle? | ### Your current environment
Does 0.11.0 suport tree attenton with eagle? Do I need to enable it manually?
### How would you like to use vllm
I want to run inference of a [specific model](put link here). I don't know how to integrate it with vllm.
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/28610 | open | [
"usage"
] | 2025-11-13T03:35:02Z | 2025-12-03T17:08:16Z | 1 | wincle |
huggingface/datasets | 7,864 | add_column and add_item erroneously(?) require new_fingerprint parameter | ### Describe the bug
Contradicting their documentation (which doesn't mention the parameter at all), both Dataset.add_column and Dataset.add_item require a new_fingerprint string. This parameter is passed directly to the dataset constructor, which has the fingerprint parameter listed as optional; is there any reason it shouldn't be optional in these methods as well?
### Steps to reproduce the bug
Reproduction steps:
1. Look at the function signature for add_column: https://github.com/huggingface/datasets/blob/17f40a318a1f8c7d33c2a4dd17934f81d14a7f57/src/datasets/arrow_dataset.py#L6078
2. Repeat for add_item: https://github.com/huggingface/datasets/blob/17f40a318a1f8c7d33c2a4dd17934f81d14a7f57/src/datasets/arrow_dataset.py#L6336
### Expected behavior
add_column and add_item should either set the fingerprint parameter to optional or include it in their docstrings
### Environment info
Not environment-dependent | https://github.com/huggingface/datasets/issues/7864 | open | [] | 2025-11-13T02:56:49Z | 2025-12-07T14:41:40Z | 2 | echthesia |
vllm-project/vllm | 28,566 | [Usage]: pd disagg scenario , I discover in the decoder , also has the prefill operation, is it normal ? | ### Your current environment
when num_computed_tokens is less than num_prompt_tokens, it will enter prefill operation
<img width="633" height="149" alt="Image" src="https://github.com/user-attachments/assets/bab96187-37c8-4ea2-ba68-9f52dda07f6b" />
and i found, num_computed_tokens is possible less than num_prompt_tokens, because num_prompt_tokens is len(block_ids) * self.block_size, event num_prompt_tokens is just equal to num_prompt_tokens, it do num_computed_tokens -= 1, why ? this cause num_computed_tokens is never equal to num_prompt_tokens
<img width="980" height="762" alt="Image" src="https://github.com/user-attachments/assets/81eb6f4f-f0db-45f8-8934-64bd8ea21988" />
### How would you like to use vllm
I want to run inference of a [specific model](put link here). I don't know how to integrate it with vllm.
### Before submitting a new issue...
- [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions. | https://github.com/vllm-project/vllm/issues/28566 | open | [
"usage"
] | 2025-11-12T16:18:53Z | 2025-11-12T16:18:53Z | 0 | yangshanjun |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.