Gemma3NPC-Filtered
Collection
A collections of Gemma3n E4B models fintuned using the pippa_filtered dataset, aimed to be a roleplaying model, though a pretty bad one
•
3 items
•
Updated
•
1
Gemma3NPC-filtered-Q8-GGUF
The "filtered" model that delivers censored general role-playing at great speed.
The Q8_0 quantized version of Gemma3NPC-filtered-float16
We trained this model as a rank-12 LoRA adapter with one epoch over pippa_filtered using a 40GB vRAM A100 in Google Colab. For this run, we employed a learning rate of 2e-5 and a total batch size of 1 and gradient accumulation steps of 16. A cosine learning rate scheduler was used with a 150-step warmup. With a gradient clipping of 0.5.
Check out our training notebook here.
Here is a graph of the Step Training Loss, saved every 5 steps:
- Downloads last month
- 3
Hardware compatibility
Log In
to view the estimation
8-bit
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support
Model tree for chimbiwide/Gemma3NPC-filtered-Q8-GGUF
Base model
chimbiwide/Gemma3NPC-filtered-float16