usage instructions

README.md

@@ -11,4 +11,23 @@ That is: we loaded Llama-13b, we applied Alpaca LoRA, expanded vocabulary, then
 This expanded tokenizer was used:  
 https://huggingface.co/Birchlabs/llama-13b-stepwise-tokenizer/blob/main/README.md
 
-You will also need the finetuned input/output embedding layers.
+You will also need the finetuned input/output embedding layers:  
+https://huggingface.co/Birchlabs/llama-13b-stepwise-embeddings/tree/main
+
+In total, you can load like this (use `evaluate.py` from our [`stepwise`](https://github.com/scottlogic-alex/qlora/tree/stepwise) branch of qlora)):  
+https://github.com/scottlogic-alex/qlora/blob/stepwise/evaluate.py#L209-L278
+
+Download `embed_tokens.pt` and `lm_head.pt` from [`Birchlabs/llama-13b-stepwise-embeddings`](https://huggingface.co/Birchlabs/llama-13b-stepwise-embeddings/tree/main), then run evaluator like so:
+
+```bash
+python -m evaluate \
+--model_name_or_path huggyllama/llama-13b \
+--base_lora_model_name_or_path chansung/alpaca-lora-13b \
+--tokenizer_model_name_or_path Birchlabs/llama-13b-stepwise-tokenizer \
+--lora_model_name_or_path Birchlabs/llama-13b-stepwise-adapter \
+--input_embedding_path embed_tokens.pt \
+--output_embedding_path lm_head.pt \
+--bf16 \
+--use_bos_token_in_prompt \
+--overrun_countermeasures False
+```