Original: https://huggingface.co/togethercomputer/GPT-JT-6B-v1

Changes

• Quantization float32 to float16

tokenizer = AutoTokenizer.from_pretrained("togethercomputer/GPT-JT-6B-v1")
model = GPTJForCausalLM.from_pretrained("ASID/GPT-JT-6B-v1-float16")