while-ai

community-loss-mask-1.5b

README

License: apache-2.0

Result, from the recipe README

Table with columns: metric, base (3 passes), as_exported, mask_honored, paired A-B [95%]
metric	base (3 passes)	`as_exported`	`mask_honored`	paired A-B [95%]
mean reply chars	173 / 168 / 154	125.7	84.1	+41.6 [+21.0, +63.7]
replies under 60 chars	0.000 / 0.018 / 0.018	0.158	0.421	-0.263 [-0.404, -0.123]
non-assistant leak (pre-registered primary)	0 / 0 / 0	0.000	0.000	+0.000

The pre-registered primary did not move: neither arm leaked non-assistant text. What moved is reply length, in the direction the mask predicts. results.json is the full comparison.

Arms in this repo

The root holds the arm the recipe README's headline number reports. Every other arm is a subfolder named after it. checkpoints/ never ships.

Table with columns: folder, arm
folder	arm
`.`	`mask_honored`: the export's own `loss_mask` applied
`as_exported`	`as_exported`: every token supervised, TRL's default

Load

python

from peft import PeftModelfrom transformers import AutoModelForCausalLM base = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-1.5B-Instruct")model = PeftModel.from_pretrained(base, "while-ai/community-loss-mask-1.5b")  # the headline armmodel = PeftModel.from_pretrained(base, "while-ai/community-loss-mask-1.5b", subfolder="as_exported")  # another arm

Reproduce

bash

git clone https://github.com/whilehq/whileai-sdk && cd whileai-sdk/recipes/community/what-trl-does-with-the-loss-maskpython run.py && modal run train_modal.py

The 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.

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Model Details

Model Provider

while-ai

Model Tree

Base

Qwen/Qwen2.5-1.5B-Instruct

Adapter

this model

Input Modalities

Text

Output Modalities