ssurface/cot-dialect-olmo3-7b-think-conditioned-sft

Olmo-3-7B-Think — single conditioned model

The alternative design: one adapter asked to serve every verbosity level by naming the level in the prompt, instead of one adapter per level. This is the comparison that motivates the per-level family.

Results

The SFT stage of the conditioned model. Numbers are reported for the GRPO stages, in the two companion repos.

Training

Stage	SFT (distillation), all levels in one corpus
Conditioning	the level is named in the prompt, not selected by adapter
Engine	HF transformers + peft
LoRA	r=16, alpha=32
Hardware	1x NVIDIA A100 80GB

Usage

Solve this using Level {N} ({Verbose|Concise|Symbolic|Shorthand|Extreme}).
Problem: {your problem}

from transformers import AutoModelForCausalLM
from peft import PeftModel

model = AutoModelForCausalLM.from_pretrained("allenai/Olmo-3-7B-Think", torch_dtype="bfloat16", device_map="auto")
model = PeftModel.from_pretrained(model, "ssurface/cot-dialect-olmo3-7b-think-conditioned-sft")

Limitations

• A design comparison, not a recommended model. The per-level adapters in this collection are the ones the results are reported on.
• Single seed.