Downloads · 30 days
51
18% of all-time downloads
LLaMa-3.2-Instruct-JankMix-v0.2-SFT-HailMary-v0.1-KTO-3B
PJMixers-Dev/LLaMa-3.2-Instruct-JankMix-v0.2-SFT-HailMary-v0.1-KTO-3B
LLaMa-3.2-Instruct-JankMix-v0.2-SFT-HailMary-v0.1-KTO-3B is a machine learning model from PJMixers-Dev. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as llama3.2.
PJMixers-Dev/LLaMa-3.2-Instruct-JankMix-v0.2-SFT-3B was further trained using KTO (with apozerounpaired loss type) using a mix of instruct, RP, and storygen datasets. I created rejected samples by using the SFT with b…
All-time downloads
288
Public
Parameters
3.2B
12.1 GB on disk
Likes
2
Public
Repo makeup
Click a slice to open those files.
.safetensors12.1 GB · 100%
Parameter types
How the weights are stored.
F322.8B · 88%
Base models
Datasets
Model card
From the Hugging Face model README
PJMixers-Dev/LLaMa-3.2-Instruct-JankMix-v0.2-SFT-3B was further trained using KTO (with apo_zero_unpaired loss type) using a mix of instruct, RP, and storygen datasets. I created rejected samples by using the SFT with bad settings (including logit bias) for every model turn.
The model was only trained at max_length=6144, and is nowhere near a full epoch as it eventually crashed. So think of this like a test of a test.
W&B Training Logs
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 21.69 |
| IFEval (0-Shot) | 65.04 |
| BBH (3-Shot) | 22.29 |
| MATH Lvl 5 (4-Shot) | 11.78 |
| GPQA (0-shot) | 2.91 |
| MuSR (0-shot) | 4.69 |
| MMLU-PRO (5-shot) | 23.42 |
- Type
- llama
- License
- llama3.2
- Languages
- en
- Created
- Oct 28, 2024
- Updated
- Oct 28, 2024