This is a speculator model designed for use with Qwen/Qwen3-VL-235B-A22B-Instruct, based on the EAGLE-3 speculative decoding algorithm. It was trained using the speculators library on a combination of the Magpie-Align/Magpie-Llama-3.1-Pro-300K-Filtered dataset and the train_sft split of the HuggingFaceH4/ultrachat_200k dataset. This model should be used with the Qwen/Qwen3-VL-235B-A22B-Instruct chat template, specifically through the /chat/completions endpoint.

Use with vLLM

bash

vllm serve Qwen/Qwen3-VL-235B-A22B-Instruct \
  -tp 8 \
  --speculative-config '{
    "model": "RedHatAI/Qwen3-VL-235B-A22B-Instruct-speculator.eagle3",
    "num_speculative_tokens": 3,
    "method": "eagle3"
  }'

Evaluations

Use cases

Use Case	Dataset	Number of Samples
Coding	HumanEval	168
Math Reasoning	gsm8k	80
Text Summarization	CNN/Daily Mail	80

Acceptance lengths

Use Case	k=1	k=2	k=3	k=4	k=5
Coding	1.85	2.49	2.88	3.13	3.24
Math Reasoning	1.85	2.54	2.96	3.25	3.24
Text Summarization	1.63	1.94	2.08	2.14	2.16

Details

Configuration

repetitions: 1
time per experiment: 10min
hardware: 8xA100
vLLM version: 0.11.2
GuideLLM version: 0.3.0

Command

bash

GUIDELLM__PREFERRED_ROUTE="chat_completions" \
guidellm benchmark \
  --target "http://localhost:8000/v1" \
  --data "RedHatAI/speculator_benchmarks" \
  --data-args '{"data_files": "HumanEval.jsonl"}' \
  --rate-type sweep \
  --max-seconds 600 \
  --output-path "Qwen235B-HumanEval.json" \

GuideLLM interface changed, so for compatibility with the latest version (v0.6.0), please use the following command:

bash

GUIDELLM__PREFERRED_ROUTE="chat_completions" \
guidellm benchmark \
  --target "http://localhost:8000/v1" \
  --data "RedHatAI/speculator_benchmarks" \
  --data-args '{"data_files": "HumanEval.jsonl"}' \
  --profile sweep \
  --max-seconds 1800 \
  --output-path "my_output.json" \
  --backend-args '{"extras": {"body": {"temperature":0.6, "top_p":0.95, "top_k":20}}}'