Add LocalAI configuration

this allows to run the model directly with LocalAI by pointing to the URL of the model file, for example:

local-ai run huggingface://fakezeta/Llama3-Aloe-8B-Alpha-ov-int8/model.yaml

model.yaml

@@ -0,0 +1,11 @@
+name: llama3-aloe
+backend: transformers
+parameters:
+  model: fakezeta/Llama3-Aloe-8B-Alpha-ov-int8
+context_size: 8192
+type: OVModelForCausalLM
+template:
+  use_tokenizer_template: true
+stopwords:
+- "<|eot_id|>"
+- "<|end_of_text|>"