vLLM Shares SkyRL Recipe and Math Benchmark Gains for Mistral Large 4

1 reports 1 reporting sources 2 hours agoNew

Understand the event

AI overview

vLLM shared a LoRA reinforcement learning recipe for Mistral Large 4 using SkyRL, reporting that it raised the model’s AIME 2024 avg@8 score from about 64% to about 75% in 75 training steps on eight NVIDIA GB200 nodes. The run used Mistral Large 4’s NVFP4 checkpoint. In the recipe, vLLM generates rollouts, while either FSDP2 or Megatron handles training. It combines rank-32 LoRA, NVFP4 quantization-aware training, FP8 KV-cache simulation, routing replay (R3), and token-level truncated importance sampling. vLLM reported a mean absolute gap of about 0.02–0.026 between FSDP2 trainer and vLLM log probabilities, with Megatron achieving similar accuracy and a comparable gap.

Generated by AI from reports · Updated 1 hours ago

Latest development10/30 09:53 (UTC+8)

vLLM reports day-0 RL results for Mistral Large 4 with SkyRL vLLM · X

Models in this story

Mistral Large 4

API prices of all models

Report timeline

Follow the reports to explore different sides of the event.

10/30

09:53
vLLM · X
vLLM reports day-0 RL results for Mistral Large 4 with SkyRL
vLLM reports that LoRA RL on Mistral Large 4’s NVFP4 checkpoint with SkyRL raises AIME 2024 avg@8 from ~64% to ~75% in 75 training steps on 8 NVIDIA GB200 nodes. vLLM generates rollouts, while FSDP2 or Megatron handles training using rank-32 LoRA, NVFP4 quantization-aware training, FP8 KV-cache simulation, routing replay (R3), and token-level truncated importance sampling. FSDP2 maintains a mean absolute trainer–vLLM logprob gap of ~0.02–0.026; Megatron achieves similar accuracy with a comparable gap.

Model releases, price changes and compute developments, with original sources. Free, confirmed by email, and easy to unsubscribe. Read a past issue

Event heat trend

There is not enough continuous observation data to plot a trend yet.