GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
amayuelas
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
amayuelas
Model Tree
Qwen/Qwen3.5-9B
Input Modalities
Output Modalities
Supported Functionality
Base Qwen3.5 essentially never commits. Across 2,880 multi-turn rollouts the 4B
base model made 3 submit calls. It proposes prolifically (~5.6
candidates/rollout), evaluates sparingly, then runs out of turns. Strict
multi-turn SUN is therefore 0% by construction: a commitment failure, not a
chemistry failure. This stage teaches the agent to close an episode.
| base | Qwen/Qwen3.5-9B |
| data | 1,119 of the dataset's 1,123 episodes, assistant-only loss |
| steps | 420 (3 epochs), global batch 8 |
| seq len | 16,384 |
| optimizer | AdamW, lr 1e-5, constant, no warmup (half the 4B's 2e-5) |
| precision | bf16 |
| parallelism | FSDP over 8 ranks + context parallel (cp=4, ulysses) |
| trainer | prime-rl |
Trained on 8×A100-40GB. Loss fell from 1.21 at the first step to ~0.8 over the final epoch.
Four episodes (3 band-gap, 1 formula) are longer than 16,384 tokens once rendered with the Qwen3.5 chat template, so they were dropped instead of truncated.
Context parallelism uses ulysses, not ring: Qwen3.5 is a hybrid with linear attention (DeltaNet) layers, and ring attention is a softmax-attention algorithm that does not apply to them.
This checkpoint has not yet been scored on the SUN benchmark used for the 4B release.
This model is trained with reasoning in the native thinking channel
(reasoning_content → <think>), preserved across tool calls. Do not
evaluate it with enable_thinking=false, which turns off the very behavior
this run trains.
Qwen3.5 is a VL-capable model class, so vLLM requires an image-processor config
even for text-only serving. preprocessor_config.json and
video_preprocessor_config.json are included; without them vLLM fails on load.