tokenizer fix

README.md

@@ -43,7 +43,7 @@ Information about location in the kaniwa chromosome: >lcl|Cp5
 
 ## Usage
 
-### Basic inference
+### Inference with DNA sequence
 
 ```python
 from peft import AutoPeftModelForCausalLM
@@ -64,16 +64,18 @@ sample = tokenizer.batch_decode(outputs, skip_special_tokens=False)[0]
 ### LoRA finetuning on a new task
 
 ```python
+from transformers import AutoTokenizer
 from trl import SFTTrainer
 from unsloth import FastLanguageModel
 
-model, tokenizer = FastLanguageModel.from_pretrained(
+model, _ = FastLanguageModel.from_pretrained(
     model_name = "monsoon-nlp/llama3-biotokenpretrain-kaniwa",
     max_seq_length = 7_000, # max 6,000 bp for AgroNT tasks
     dtype = None,
     load_in_4bit = True,
     resize_model_vocab=128260, # includes biotokens
 )
+tokenizer = AutoTokenizer.from_pretrained("monsoon-nlp/llama3-biotokenpretrain-kaniwa")
 tokenizer.pad_token = tokenizer.eos_token # pad fix
 
 trainer = SFTTrainer(