---
library_name: peft
base_model: Qwen/Qwen3-8B
tags:
- lora
- transformers
pipeline_tag: text-generation
---

# qwen3-8b-navigation-lora-persistent

Anonymous supplementary release for a double-blind workshop submission. This is
one of four LoRA adapters (rule_diagnosis / navigation task family x
persistent/stateless training regime), fine-tuned on the navigation agentic task (graph exploration with a per-turn tool-call budget). It is the
second-family generalization arm alongside the primary Opaque Knapsack result
(see the sibling Qwen3-8B knapsack release).

- **Base model:** [Qwen/Qwen3-8B](https://huggingface.co/Qwen/Qwen3-8B)
- **Training regime:** persistent (trained with a persistent Python interpreter runtime (state carries over across agent turns))
- **Seed:** 3407

## Training configuration

Fine-tuned with [Axolotl](https://github.com/axolotl-ai-cloud/axolotl), LoRA
adapter, 4-bit NF4 quantized base:

| Hyperparameter | Value |
|---|---|
| lora_r | 64 |
| lora_alpha | 128 |
| lora_dropout | 0.05 |
| lora_target_modules | q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj |
| learning_rate | 1e-4 |
| lr_scheduler | cosine |
| optimizer | adamw_torch |
| epochs | 3.0 |
| micro_batch_size | 1 |
| gradient_accumulation_steps | 16 |
| sequence_len | 16384 |
| sample_packing | false |
| seed | 3407 |
| training data | paired traces for the "persistent" regime on navigation, see paper Appendix for pairing/filtering procedure |

## Provenance

Released anonymously alongside a NeurIPS workshop submission for
reproducibility review. Non-anonymous release (paper citation, full code, full
training traces) will follow after the review process concludes.