Create mamba.py

models/mamba.py

@@ -0,0 +1,16 @@
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+model_name = "s3nh/mamba-gpt-3b-v3-GGML"
+
+def load():
+  global model
+  global tokenizer
+  
+  model = AutoModelForCausalLM.from_pretrained(model_name)
+  tokenizer = AutoTokenizer.from_pretrained(model_name)
+
+def generate(input_text):
+  input_ids = tokenizer.encode(input_text, return_tensors="pt")
+  output_ids = model.generate(input_ids, num_beams=5, no_repeat_ngram_size=2)
+  
+  return tokenizer.decode(output_ids[0], skip_special_tokens=True)