Llama-70B-God-Tier

Full merged Transformers checkpoint (sharded safetensors) created by merging a LoRA adapter into meta-llama/Llama-3.3-70B-Instruct.

Model Details

• Type: merged fine-tune (full weights)
• Base model: meta-llama/Llama-3.3-70B-Instruct
• Adapter repo: https://huggingface.co/Daga2001/Llama-70B-God-Tier-Adapter
• Dataset: tatsu-lab/alpaca
• Training samples: 1000
• Max sequence length: 512
• Packing: False
• LoRA (merged): r=16, alpha=32, dropout=0.05, target=attn

Intended Use

This repo is intended for Transformers users.

Usage (Transformers)

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "Daga2001/Llama-70B-God-Tier"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")

LM Studio

LM Studio typically expects GGUF files. See the *-GGUF repo if provided.

License / Attribution

Use is subject to the base model license and terms.