Update README.md

README.md

@@ -5,9 +5,9 @@ tags: [custom_generate]
 
 
 ## Description
-Test repo to experiment with calling `generate` from the hub. It is a simplified implementation of greedy decoding.
+Example repository used to document `generate` from the hub.
 
-⚠️ this recipe has an impossible requirement and is meant to crash. If you try to run it, you should see something like
+⚠️ this custom generation method has an impossible requirement and is meant to crash. If you try to run it, you should see something like
 ```
 ValueError: Missing requirements for joaogante/test_generate_from_hub_bad_requirements:
 foo (installed: None)
@@ -27,3 +27,18 @@ Most models. More specifically, any `transformer` LLM/VLM trained for causal lan
 ## Output Type changes
 (none)
 
+## Example usage
+
+```py
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+# `generate` with `custom_generate` -> `generate` uses custom code
+# note: calling the custom method prints "✨ using a custom generation method ✨"
+tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-0.5B-Instruct")
+model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-0.5B-Instruct", device_map="auto")
+
+inputs = tokenizer(["The quick brown"], return_tensors="pt").to(model.device)
+gen_out = model.generate(**inputs, custom_generate="transformers-community/custom_generate_bad_requirements", trust_remote_code=True)
+print(tokenizer.batch_decode(gen_out, skip_special_tokens=True))
+```
+