Downloads · 30 days
0
LLM2026_DPO_SFT19_v17
makotonlo/LLM2026_DPO_SFT19_v17
LLM2026_DPO_SFT19_v17 is a machine learning model from makotonlo. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
This model is the latest version of the Silent Expert, evolved from the high-performing SFT model makotonlo/LLM2026SFTfinalv197B (v19) (Score: 0.767). It has been strictly fine-tuned using DPO with Beta 0.7 to complet…
Access
Public
Updated Mar 1, 2026
Repo size
1.3 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors1.3 GB · 99%
At a glance
- Access
- Public
- Created
- Mar 1, 2026
- Updated
- Mar 1, 2026
- SHA
- c0e75704
Base models
Model card
From the Hugging Face model README
LLM2026_DPO_SFT19_v17 (Silent Expert v17)
This model is the latest version of the Silent Expert, evolved from the high-performing SFT model makotonlo/LLM2026_SFT_finalv19_7B (v19) (Score: 0.767). It has been strictly fine-tuned using DPO with Beta 0.7 to completely eliminate conversational noise.
🎯 Optimization Goal (Absolute Silence)
Ensures the model outputs ONLY raw data (JSON, XML, YAML, CSV) without:
- Preambles (e.g., "Certainly!", "To convert this...")
- Markdown backticks (
json ...) - Post-output explanations (e.g., "Note: ...")
🛠 Configuration (v17 Final Specs)
- Base Intelligence: v19 (0.767 accuracy)
- Method: DPO (Direct Preference Optimization)
- Learning Rate: 1e-05
- Beta: 0.7 (Maximum penalty for any non-data character)
- Epochs: 3.0
- LoRA Config: r=128, alpha=128
⚠️ Important: Usage Note
Please use the ChatML template for inference. The model is optimized to start its response directly with data-starting characters.
- Created
- Mar 1, 2026
- Updated
- Mar 1, 2026