fix vocab size

```python
from transformers import AutoTokenizer
testtokenizer=AutoTokenizer.from_pretrained("LeoLM/leo-mistral-hessianai-7b-chat")
len(testtokenizer)
# 32002
```

Leads to e.g. VLLM error:
`TypeError: argument 'tokens': 'NoneType' object cannot be converted to 'PyString'`
(see [here](https://github.com/vllm-project/vllm/issues/516#issuecomment-1657507293 ))

config.json

@@ -21,5 +21,5 @@
   "torch_dtype": "float16",
   "transformers_version": "4.34.0",
   "use_cache": true,
-  "vocab_size": 32128
+  "vocab_size": 32002
 }