Chatmlification meme

Base model: Captain Eris Violet-v0.420

Step 1: Tokens swapped to ChatML "appropriately" in configs. (This already happened)

Step 2: Barycentric based embedding swap applied with token surgery. (You are currently here) Example notebook on how to accomplish this

To-do.

Step 3: Train for a single epoch on additional instruction data for healing of the network.

Downloads last month: 2

Safetensors

Model size

12.2B params

Tensor type

BF16

Inference Providers NEW

This model is not currently available via any of the supported third-party Inference Providers, and HF Inference API was unable to determine this model's library.