Downloads · 30 days
0
math-model-targeted-v2-l45-g4b-lora
Yuivdldk/math-model-targeted-v2-l45-g4b-lora
math-model-targeted-v2-l45-g4b-lora is a machine learning model from Yuivdldk. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
This repository contains the LoRA adapter weights for the targeted v2 external teacher arm trained on Gemma-3-4B-it.
Access
Public
Updated Jul 27, 2026
Repo size
153 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors119 MB · 78%
At a glance
- Access
- Public
- Created
- Jul 27, 2026
- Updated
- Jul 27, 2026
- SHA
- 159a666b
Model card
From the Hugging Face model README
Targeted v2 LoRA Adapter (Gemma-3-4B-it)
This repository contains the LoRA adapter weights for the targeted v2 external teacher arm trained on Gemma-3-4B-it.
Overview & Performance
- Arm Name: targeted v2 (265 verified teacher solutions + 265 replay solutions)
- Ledger Reference:
AI_AGENT_WORKFLOW.md§1 Row 47 (Line 47) - Evaluation Delta: −4.45pp vs Baseline (W/L 169/282)
- Evaluation Dataset Repo:
Yuivdldk/mathvl-l45-targetedv2-g4b-mn4096bi-n2538
Training Recipe & Hyperparameters
- Base Model:
google/gemma-3-4b-it - Base Revision:
093f9f388b31de276ce2de164bdc2081324b9767 - Learning Rate:
1e-4 - Epochs:
1.0 - Seed:
42 - LoRA Configuration: Target 238 modules in language model layers
- Training Set:
targeted_sft_v2.jsonl(sha256:c8aa7aae1c55db81432001fb08cad0b4ac3c756f7d509e3f91e78ac07d09c672)
- Created
- Jul 27, 2026
- Updated
- Jul 27, 2026