icedduck/lab1-armA

Countdown Arm A: calculator tool + ask-an-expert with a budget; CD-4 pass@1 .729 alone, .979 with the expert (report ยง3.10).

Part of LLM from scratch, to the aha moment, to a coding agent: fifteen experiments on four RTX 4090s. Code, report, the evaluation protocol and the scripts that produced this checkpoint: https://github.com/bethehand/lab1-llm-scratch-to-aha-moment-to-coding-agent

Evaluate it yourself from the repository root, for example python3 bench_observation_price.py --hf-dir icedduck/lab1-armA for the coding-line checkpoints. Weights are derived from Qwen/Qwen2.5-1.5B and remain under the Qwen license.

Downloads last month
12
Safetensors
Model size
2B params
Tensor type
BF16
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for icedduck/lab1-armA

Finetuned
(467)
this model