Update README.md

README.md

@@ -13,6 +13,16 @@ A continued pretrained version of unsloth/Qwen2.5-7B model using unsloth's low r
 
 For pretraining, posts from [SubMaroon/DTF_comments_Responses_Counts](https://huggingface.co/datasets/SubMaroon/DTF_Comments_Responses_Counts) were selected, deduplicated by simple `df.unique` and filtered by length of 1000 < x < 128000 tokens.
 
+The model is trained for text completion and not optimized for conversational usage. If you want to use it anyway, here is usage example:
+
+```
+import transformers
+
+pipe = transformers.pipeline('text-generation', model='./Qwen2.5-7B-DTF-pretrain/', device=0)
+print(pipe('Кто такой карасик?', max_new_tokens=45))
+# >>>  [{'generated_text': 'Кто такой карасик? Это неизвестный, но очень опасный и сильный враг, который может убить любого героя. Он может быть в любой момент, в любом месте, и его невозможно предугадать.'}]
+```
+
 LoRA hyperparameters:
 
 ```