GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
while-ai
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
while-ai
Model Tree
Qwen/Qwen2.5-Coder-1.5B-Instruct
Input Modalities
Output Modalities
Supported Functionality
2026-09-21, one H100, 79 held-out tasks in six families the training never showed, 4 rollouts a task, 40 GRPO steps per trained arm.
| Cell | harness | model | pass@1 | 95% CI |
|---|---|---|---|---|
| neither | 00_baseline | base | 0.00 | [0.00, 0.01] |
| harness | 01_skills | base | 0.09 | [0.05, 0.13] |
| weights | 00_baseline | trained | 0.03 | [0.01, 0.04] |
| both | 01_skills | trained | 0.09 | [0.06, 0.12] |
Picking the harness on the holdout moved the base +0.08 [+0.05, +0.12]. Training the weights under the bare harness moved it +0.03. Training under the chosen harness landed where the harness alone did on pass@1 and raised pass@4 from 0.09 to 0.30. One training seed per arm, so the trained pairs are unresolved. The recipe README has the full attribution table and both hack scans.
The root holds the arm the recipe README's headline number reports. Every other arm is a subfolder named after it. checkpoints/ never ships.
| folder | arm |
|---|---|
. | both: trained under the 01_skills harness, 2026-09-21 run |
weights | weights: trained under the bare 00_baseline harness, 2026-09-21 run |
python
from peft import PeftModelfrom transformers import AutoModelForCausalLM base = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-Coder-1.5B-Instruct")model = PeftModel.from_pretrained(base, "while-ai/paper-harness-and-weights-1.5b") # the headline armmodel = PeftModel.from_pretrained(base, "while-ai/paper-harness-and-weights-1.5b", subfolder="weights") # another armbash
git clone https://github.com/whilehq/whileai-sdk && cd whileai-sdk/recipes/papers/harness-and-weightspython recipe.pyThe recipe README pins the seed, the library versions and the GPU, and its Checks table says what the eval verified. Read the Learned section before quoting a number from this card.
from peft import PeftModelfrom transformers import AutoModelForCausalLM base = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-Coder-1.5B-Instruct")model = PeftModel.from_pretrained(base, "while-ai/paper-harness-and-weights-1.5b") # the headline armmodel = PeftModel.from_pretrained(base, "while-ai/paper-harness-and-weights-1.5b", subfolder="weights") # another armgit clone https://github.com/whilehq/whileai-sdk && cd whileai-sdk/recipes/papers/harness-and-weightspython recipe.py