[MoE Refactor][15/N] Apply Refactor to Fp8 (#31415)

This commit is contained in:
Robert Shaw
2026-01-07 19:42:33 -05:00
committed by GitHub
parent ffc0a2798b
commit 5dcd7ef1f2
38 changed files with 1439 additions and 1528 deletions
@@ -0,0 +1,5 @@
model_name: "nvidia/Llama-4-Scout-17B-16E-Instruct-FP8"
accuracy_threshold: 0.92
num_questions: 1319
num_fewshot: 5
server_args: "--enforce-eager --max-model-len 8192 --data-parallel-size 2 --enable-expert-parallel"
@@ -0,0 +1,8 @@
model_name: "Qwen/Qwen3-Coder-30B-A3B-Instruct-FP8"
accuracy_threshold: 0.88
num_questions: 1319
num_fewshot: 5
server_args: "--enforce-eager --max-model-len 8192 --data-parallel-size 2 --enable-expert-parallel --all2all-backend deepep_high_throughput"
env:
VLLM_USE_DEEP_GEMM: "1"
VLLM_USE_DEEP_GEMM_MOE: "1"
@@ -0,0 +1,9 @@
model_name: "Qwen/Qwen3-Coder-30B-A3B-Instruct-FP8"
accuracy_threshold: 0.88
num_questions: 1319
num_fewshot: 5
server_args: "--enforce-eager --max-model-len 8192 --data-parallel-size 2 --enable-expert-parallel --all2all-backend deepep_low_latency --disable-uvicorn-access-log"
env:
VLLM_USE_DEEP_GEMM: "1"
VLLM_USE_DEEP_GEMM_MOE: "1"
VLLM_USE_DEEP_GEMM_E8M0: "0"
@@ -0,0 +1,8 @@
model_name: "Qwen/Qwen3-Coder-30B-A3B-Instruct-FP8"
accuracy_threshold: 0.88
num_questions: 1319
num_fewshot: 5
server_args: "--enforce-eager --max-model-len 8192 --data-parallel-size 2 --enable-expert-parallel"
env:
VLLM_USE_DEEP_GEMM: "1"
VLLM_USE_DEEP_GEMM_MOE: "1"
@@ -0,0 +1,8 @@
model_name: "RedHatAI/Qwen3-30B-A3B-FP8-block"
accuracy_threshold: 0.85
num_questions: 1319
num_fewshot: 5
server_args: "--enforce-eager --max-model-len 8192 --data-parallel-size 2 --enable-expert-parallel --all2all-backend deepep_high_throughput"
env:
VLLM_USE_DEEP_GEMM: "1"
VLLM_USE_DEEP_GEMM_MOE: "1"
@@ -0,0 +1,9 @@
model_name: "RedHatAI/Qwen3-30B-A3B-FP8-block"
accuracy_threshold: 0.85
num_questions: 1319
num_fewshot: 5
server_args: "--enforce-eager --max-model-len 8192 --data-parallel-size 2 --enable-expert-parallel --all2all-backend deepep_low_latency --disable-uvicorn-access-log"
env:
VLLM_USE_DEEP_GEMM: "1"
VLLM_USE_DEEP_GEMM_MOE: "1"
VLLM_USE_DEEP_GEMM_E8M0: "0"
@@ -0,0 +1,8 @@
model_name: "RedHatAI/Qwen3-30B-A3B-FP8-block"
accuracy_threshold: 0.85
num_questions: 1319
num_fewshot: 5
server_args: "--enforce-eager --max-model-len 8192 --data-parallel-size 2 --enable-expert-parallel"
env:
VLLM_USE_DEEP_GEMM: "1"
VLLM_USE_DEEP_GEMM_MOE: "1"
@@ -0,0 +1,8 @@
model_name: "RedHatAI/Qwen3-30B-A3B-NVFP4"
accuracy_threshold: 0.88
num_questions: 1319
num_fewshot: 5
server_args: "--enforce-eager --max-model-len 8192 --data-parallel-size 2 --enable-expert-parallel"
env:
VLLM_USE_FLASHINFER_MOE_FP4: "1"
VLLM_FLASHINFER_MOE_BACKEND: "throughput"
@@ -0,0 +1,8 @@
Qwen3-30B-A3B-Fp8-AutoFp8-deepgemm-deepep-ht.yaml
Qwen3-30B-A3B-Fp8-AutoFp8-deepgemm-deepep-ll.yaml
Qwen3-30B-A3B-Fp8-AutoFp8-deepgemm.yaml
Qwen3-30B-A3B-Fp8-CT-Block-deepgemm-deepep-ht.yaml
Qwen3-30B-A3B-Fp8-CT-Block-deepgemm-deepep-ll.yaml
Qwen3-30B-A3B-Fp8-CT-Block-deepgemm.yaml
Qwen3-30B-A3B-NvFp4-CT-fi-cutlass.yaml
Qwen3-30B-A3B-NvFp4-ModelOpt-fi-cutlass-fi-a2av.yaml
@@ -0,0 +1,5 @@
model_name: "RedHatAI/Llama-4-Scout-17B-16E-Instruct-FP8-dynamic"
accuracy_threshold: 0.92
num_questions: 1319
num_fewshot: 5
server_args: "--enforce-eager --max-model-len 8192 --tensor-parallel-size 2"
@@ -1,3 +1,4 @@
Llama-4-Scout-Fp8-CT-vllm-cutlass.yaml
Llama-4-Scout-Fp8-ModelOpt-fi-trtllm.yaml
Qwen3-30B-A3B-Fp8-AutoFp8-fi-trtllm.yaml
Qwen3-30B-A3B-NvFp4-CT-vllm-cutlass.yaml
@@ -5,7 +5,7 @@ Qwen3-30B-A3B-Fp8-AutoFp8-marlin.yaml
Qwen3-30B-A3B-Fp8-AutoFp8-triton.yaml
Qwen3-30B-A3B-Fp8-CT-Block-deepgemm.yaml
Qwen3-30B-A3B-Fp8-CT-Block-marlin.yaml
Qwen3-30B-A3B-Fp8-CT-Block-vllm-cutlass.yaml
Qwen3-30B-A3B-Fp8-CT-Block-triton.yaml
Qwen3-30B-A3B-Fp8-CT-Channel-marlin.yaml
Qwen3-30B-A3B-Fp8-CT-Channel-vllm-cutlass.yaml
Llama-4-Scout-Fp8-ModelOpt-fi-cutlass.yaml
@@ -61,6 +61,7 @@ def test_gsm8k_correctness(config_filename):
server_args.extend(
[
"--trust-remote-code",
"--disable-uvicorn-access-log",
]
)