commit

examples/model_configs/vllm_model_config.yaml

@@ -1,6 +1,14 @@
 model:
   base_params:
-    model_args: "pretrained=NousResearch/DeepHermes-3-Llama-3-8B-Preview,revision=main,dtype=bfloat16,max_model_length=32000" # pretrained=model_name,trust_remote_code=boolean,revision=revision_to_use,model_parallel=True ...
+    model_args: "pretrained=HuggingFaceTB/SmolLM-1.7B,revision=main,dtype=bfloat16" # pretrained=model_name,trust_remote_code=boolean,revision=revision_to_use,model_parallel=True ...
   generation:
     temperature: 0.3
-    max_new_tokens: 12000
+    repetition_penalty: 1.0
+    frequency_penalty: 0.0
+    presence_penalty: 0.0
+    seed: 42
+    top_k: -1
+    min_p: 0.0
+    top_p: 0.9
+    max_new_tokens: 256
+    stop_tokens: ["<EOS>", "<PAD>"]