stanford_alpaca
stanford_alpaca copied to clipboard
tatsu-lab
Code and documentation to train Stanford's Alpaca models, and generate the data.
https://fairscale.readthedocs.io/en/stable/_modules/fairscale/nn/data_parallel/fully_sharded_data_parallel.html#FullyShardedDataParallel.consolidate_shard_weights
./main -m ./models/ggml-vicuna-13b-4bit.bin --color -f ./prompts/alpaca.txt -ins -b 256 --top_k 10000 --temp 0.2 --repeat_penalty 1 -t 7 Below is the conversation, which is good as far as it goes. If...
As title, I found that the content of llama-Xb-hf/tokenizer_config.json is like the following, ``` {"bos_token": "", "eos_token": "", "model_max_length": 1000000000000000019884624838656, "tokenizer_class": "LLaMATokenizer", "unk_token": ""} ``` How did your team modify...
Python: 3.9.5 Ubuntu 20.04.6 LTS ``` WARNING:torch.distributed.run: ***************************************** Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the...
Bumped into this whilst looking at the file. Surely this must have been a typo whilst assembling it?
I used to successfully run alpaca on my own dataset last year. But I rerun the train.py recently, for both alpaca dataset and my own dataset, they both failed for...
When I was training, this error occurred
