Modelpublic

runtime-contracts/qwen3-8b-knapsack-lora-stateless-seed777

sourceHugging Faceapache-2.0updated 3d agoView on Hugging Face

0likes14downloads

Model Card

qwen3-8b-knapsack-lora-stateless-seed777

Supplementary release for the paper Evaluating Agents Across Runtime Contracts: When Mismatch Costs Efficiency or Quality (IAEval 2026, the NeurIPS 2026 Workshop on Evaluation of Interactive Agents). This is one of six LoRA adapters (persistent/stateless training regime x 3 seeds) fine-tuned on the Opaque Knapsack agentic task.

—Base model: Qwen/Qwen3-8B
—Training regime: stateless (trained with a stateless Python interpreter runtime (interpreter state is reset every agent turn))
—Seed: 777

Training configuration

Fine-tuned with Axolotl 0.13.2, LoRA adapter, 4-bit NF4 quantized base:

Hyperparameter	Value
lora_r	64
lora_alpha	128
lora_dropout	0.05
loratargetmodules	qproj, kproj, vproj, oproj, gateproj, upproj, down_proj
learning_rate	1e-4
lr_scheduler	cosine
optimizer	adamw_torch
epochs	3.0
microbatchsize	1
gradientaccumulationsteps	16
sequence_len	16384
sample_packing	false
seed	777
training data	paired traces for the "stateless" regime (see paper Appendix for pairing/filtering procedure)

Provenance

Released alongside the paper Evaluating Agents Across Runtime Contracts: When Mismatch Costs Efficiency or Quality (IAEval 2026, the NeurIPS 2026 Workshop on Evaluation of Interactive Agents), to reproduce its reported results.

License

Apache License 2.0. The base model (Qwen3-8B) is also released under Apache 2.0.