metadata

base_model: unsloth/llama-3-8b-bnb-4bit
language:
  - en
license: apache-2.0
tags:
  - text-generation-inference
  - transformers
  - unsloth
  - llama
  - trl
  - sft

Still in Beta!!!

This model will be used for the base model in J.O.S.I.E.v4o and furhter trained.

Uploaded model

Developed by: Isaak-Carter
License: apache-2.0
Finetuned from model : Isaak-Carter/JOSIEv4o-8b-stage1-beta1
Dataset used : Isaak-Carter/j.o.s.i.e.v4.0.1o

This llama model was trained 2x faster with Unsloth and Huggingface's TRL library.

  "per_device_train_batch_size": 2,
  "gradient_accumulation_steps": 8,
  "max_steps": 300,
  "learning_rate": 2e-4,
  "optim": "adamw_8bit",
  "weight_decay": 0.01,
  "lr_scheduler_type": "cosine"

trained on this Prompt format

"""<|begin_of_text|>system
You are J.O.S.I.E. which is an acronym for "Just an Outstandingly Smart Intelligent Entity", a private and super-intelligent AI assistant, created by Gökdeniz Gülmez.
<|begin_of_text|>main user "Gökdeniz Gülmez"
{{ .Prompt }}
<|begin_of_text|>josie
{{ .Response }}<|end_of_text|>"""