remote-vllm provider does not support logprobs and does not pass TestInference::test_completion_logprobs
#1073
Closed
1 of 2 tasks
Labels
bug
Something isn't working
Comments
bbrowning
added a commit
to bbrowning/llama-stack
that referenced
this issue
Feb 13, 2025
The remote-vllm provider was not passing logprobs options from CompletionRequest or ChatCompletionRequests through to the OpenAI client parameters. I manually verified this, as well as observed this provider failing `TestInference::test_completion_logprobs`. This fixes that by passing the `logprobs.top_k` value through to the parameters we pass into the OpenAI client. Additionally, this fixes a bug in `test_text_inference.py` where it mistakenly assumed chunk.delta were of type `ContentDelta` for completion requests. The deltas are of type `ContentDelta` for chat completion requests, but for basic completion requests the deltas are of type string. This test was likely failing for other providers that did properly support logprobs because of this latter issue in the test, which was hit while fixing the above issue with the remote-vllm provider. Fixes meta-llama#1073 Signed-off-by: Ben Browning <[email protected]>
bbrowning
changed the title
TestInference::test_completion_logprobsTestInference::test_completion_logprobs
bbrowning
changed the title
TestInference::test_completion_logprobsTestInference::test_completion_logprobs
terrytangyuan
pushed a commit
that referenced
this issue
Feb 13, 2025
# What does this PR do? The remote-vllm provider was not passing logprobs options from CompletionRequest or ChatCompletionRequests through to the OpenAI client parameters. I manually verified this, as well as observed this provider failing `TestInference::test_completion_logprobs`. This was filed as issue #1073. This fixes that by passing the `logprobs.top_k` value through to the parameters we pass into the OpenAI client. Additionally, this fixes a bug in `test_text_inference.py` where it mistakenly assumed chunk.delta were of type `ContentDelta` for completion requests. The deltas are of type `ContentDelta` for chat completion requests, but for basic completion requests the deltas are of type string. This test was likely failing for other providers that did properly support logprobs because of this latter issue in the test, which was hit while fixing the above issue with the remote-vllm provider. (Closes #1073) ## Test Plan First, you need a vllm running. I ran one locally like this: ``` vllm serve meta-llama/Llama-3.2-3B-Instruct --port 8001 --enable-auto-tool-choice --tool-call-parser llama3_json ``` Next, run test_text_inference.py against this vllm using the remote vllm provider like this: ``` VLLM_URL="http://localhost:8001/v1" python -m pytest -s -v llama_stack/providers/tests/inference/test_text_inference.py --providers "inference=vllm_remote" ``` Before my change, the test failed with this error: ``` llama_stack/providers/tests/inference/test_text_inference.py:155: in test_completion_logprobs assert 1 <= len(response.logprobs) <= 5 E TypeError: object of type 'NoneType' has no len() ``` After my change, the test passes. [//]: # (## Documentation) Signed-off-by: Ben Browning <[email protected]>
yanxi0830
pushed a commit
that referenced
this issue
Feb 13, 2025
# What does this PR do? The remote-vllm provider was not passing logprobs options from CompletionRequest or ChatCompletionRequests through to the OpenAI client parameters. I manually verified this, as well as observed this provider failing `TestInference::test_completion_logprobs`. This was filed as issue #1073. This fixes that by passing the `logprobs.top_k` value through to the parameters we pass into the OpenAI client. Additionally, this fixes a bug in `test_text_inference.py` where it mistakenly assumed chunk.delta were of type `ContentDelta` for completion requests. The deltas are of type `ContentDelta` for chat completion requests, but for basic completion requests the deltas are of type string. This test was likely failing for other providers that did properly support logprobs because of this latter issue in the test, which was hit while fixing the above issue with the remote-vllm provider. (Closes #1073) ## Test Plan First, you need a vllm running. I ran one locally like this: ``` vllm serve meta-llama/Llama-3.2-3B-Instruct --port 8001 --enable-auto-tool-choice --tool-call-parser llama3_json ``` Next, run test_text_inference.py against this vllm using the remote vllm provider like this: ``` VLLM_URL="http://localhost:8001/v1" python -m pytest -s -v llama_stack/providers/tests/inference/test_text_inference.py --providers "inference=vllm_remote" ``` Before my change, the test failed with this error: ``` llama_stack/providers/tests/inference/test_text_inference.py:155: in test_completion_logprobs assert 1 <= len(response.logprobs) <= 5 E TypeError: object of type 'NoneType' has no len() ``` After my change, the test passes. [//]: # (## Documentation) Signed-off-by: Ben Browning <[email protected]>
franciscojavierarceo
pushed a commit
to franciscojavierarceo/llama-stack
that referenced
this issue
Feb 14, 2025
# What does this PR do? The remote-vllm provider was not passing logprobs options from CompletionRequest or ChatCompletionRequests through to the OpenAI client parameters. I manually verified this, as well as observed this provider failing `TestInference::test_completion_logprobs`. This was filed as issue meta-llama#1073. This fixes that by passing the `logprobs.top_k` value through to the parameters we pass into the OpenAI client. Additionally, this fixes a bug in `test_text_inference.py` where it mistakenly assumed chunk.delta were of type `ContentDelta` for completion requests. The deltas are of type `ContentDelta` for chat completion requests, but for basic completion requests the deltas are of type string. This test was likely failing for other providers that did properly support logprobs because of this latter issue in the test, which was hit while fixing the above issue with the remote-vllm provider. (Closes meta-llama#1073) ## Test Plan First, you need a vllm running. I ran one locally like this: ``` vllm serve meta-llama/Llama-3.2-3B-Instruct --port 8001 --enable-auto-tool-choice --tool-call-parser llama3_json ``` Next, run test_text_inference.py against this vllm using the remote vllm provider like this: ``` VLLM_URL="http://localhost:8001/v1" python -m pytest -s -v llama_stack/providers/tests/inference/test_text_inference.py --providers "inference=vllm_remote" ``` Before my change, the test failed with this error: ``` llama_stack/providers/tests/inference/test_text_inference.py:155: in test_completion_logprobs assert 1 <= len(response.logprobs) <= 5 E TypeError: object of type 'NoneType' has no len() ``` After my change, the test passes. [//]: # (## Documentation) Signed-off-by: Ben Browning <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
System Info
Collecting environment information...
PyTorch version: 2.6.0+cpu
Is debug build: False
CUDA used to build PyTorch: None
ROCM used to build PyTorch: N/A
OS: Fedora Linux 40 (Workstation Edition) (x86_64)
GCC version: (GCC) 14.2.1 20240912 (Red Hat 14.2.1-3)
Clang version: Could not collect
CMake version: Could not collect
Libc version: glibc-2.39
Python version: 3.12.8 (main, Dec 6 2024, 00:00:00) [GCC 14.2.1 20240912 (Red Hat 14.2.1-3)] (64-bit runtime)
Python platform: Linux-6.12.11-100.fc40.x86_64-x86_64-with-glibc2.39
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:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 39 bits physical, 48 bits virtual
Byte Order: Little Endian
CPU(s): 8
On-line CPU(s) list: 0-7
Vendor ID: GenuineIntel
Model name: 11th Gen Intel(R) Core(TM) i7-1185G7 @ 3.00GHz
CPU family: 6
Model: 140
Thread(s) per core: 2
Core(s) per socket: 4
Socket(s): 1
Stepping: 1
CPU(s) scaling MHz: 50%
CPU max MHz: 4800.0000
CPU min MHz: 400.0000
BogoMIPS: 5990.40
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 sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l2 cdp_l2 ssbd ibrs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves split_lock_detect user_shstk dtherm ida arat pln pts hwp hwp_notify hwp_act_window hwp_epp hwp_pkg_req vnmi avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq rdpid movdiri movdir64b fsrm avx512_vp2intersect md_clear ibt flush_l1d arch_capabilities
Virtualization: VT-x
L1d cache: 192 KiB (4 instances)
L1i cache: 128 KiB (4 instances)
L2 cache: 5 MiB (4 instances)
L3 cache: 12 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-7
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: 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 SW loop, KVM SW loop
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected
Versions of relevant libraries:
[pip3] numpy==1.26.4
[pip3] torch==2.6.0+cpu
[pip3] torchvision==0.21.0+cpu
[conda] Could not collect
Information
🐛 Describe the bug
The remote-vllm provider is failing the test
llama_stack/providers/tests/inference/test_text_inference.py::TestInference::test_completion_logprobsand generally does not support requests for logprobs. When the client passes a validLogProbConfigin their inference request, the remote-vllm provider ignores that and doesn't actually pass the proper logprobs options through to its underlying OpenAI client.Error logs
Expected behavior
The remote-vllm provider should pass
test_completion_logprobsand should generally work with logprobs requests.The text was updated successfully, but these errors were encountered: