Downloads · 30 days
5
10% of all-time downloads
ps4mas-sft-lora-ceod-gate-v1
yinita/ps4mas-sft-lora-ceod-gate-v1
ps4mas-sft-lora-ceod-gate-v1 is a machine learning model from yinita. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for peft.
Teacher-gated SFT LoRA on top of the x5 mixed full-FT bottom-line. Best remaining SFT LoRA checkpoint on disk (early sftfrom / sft1e-4 adapters were pruned).
All-time downloads
52
Public
Repo size
486 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors466 MB · 96%
At a glance
- Library
- peft
- Access
- Public
- Created
- Jul 18, 2026
- Updated
- Jul 18, 2026
- SHA
- 49e0d5a2
Base models
Model card
From the Hugging Face model README
PS4MAS SFT LoRA — ceod_gate_lr_sft_v1 (best available)
Teacher-gated SFT LoRA on top of the x5 mixed full-FT bottom-line. Best remaining SFT LoRA checkpoint on disk (early sft_from_* / sft_1e-4_* adapters were pruned).
- Local run:
results/ceod/ceod_gate_lr_sft_v1/final_lora - HF repo:
yinita/ps4mas-sft-lora-ceod-gate-v1 - Base (required): yinita/ps4mas-sft-x5-mixed-ep3
- LoRA: r=64, alpha=128, dropout=0.05
- Eval:
tiny_eval— 50×4, GPT-OSS-120B judge
tiny_eval results
| Model | Method | Overall | Single | Central | Hier | Debate | avg_ask | zero_tool% |
|---|---|---|---|---|---|---|---|---|
| baseline_9b (Qwen3.5-9B) | raw | 3.386 | 2.140 | 3.613 | 3.508 | 4.281 | 1.74 | 90.0 |
| x5_mixed_ep1 (this / sibling) | SFT full-FT | 3.679 | 2.387 | 4.101 | 3.821 | 4.407 | 1.95 | 48.0 |
| x5_mixed_ep3 (bottom-line) | SFT full-FT | 3.622 | 2.267 | 3.974 | 3.801 | 4.448 | 1.93 | 58.0 |
| x5_single_ep3 | SFT full-FT | 3.646 | 2.600 | 3.821 | 3.881 | 4.281 | 2.13 | 6.0 |
| ceod_gate_lr_sft_v1 | SFT LoRA | 3.602 | 2.326 | 4.094 | 3.747 | 4.241 | 1.89 | 58.0 |
| low400_ep3 (weights not on disk) | SFT full-FT | 3.682 | 2.546 | 3.854 | 3.860 | 4.467 | 1.86 | 66.0 |
Usage
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer
base = "yinita/ps4mas-sft-x5-mixed-ep3"
adapter = "yinita/ps4mas-sft-lora-ceod-gate-v1"
model = AutoModelForCausalLM.from_pretrained(base, torch_dtype="auto", device_map="auto")
model = PeftModel.from_pretrained(model, adapter)
tokenizer = AutoTokenizer.from_pretrained(adapter)
Citation / project
PS4MAS — Personalized Safety Multi-Agent Systems (internal research checkpoint).
- Library
- peft
- Created
- Jul 18, 2026
- Updated
- Jul 18, 2026