update falcon metadata (#404)

A couple of small changes for consistency.

Also in the lord of the rings it was eagles, not falcons.

falcon/falcon3-10B-trt-llm-spec-dec/config.yaml

@@ -1,20 +1,16 @@
 model_metadata:
-  tags:
-    - openai-compatible
   example_model_input:
     frequency_penalty: 1
     max_tokens: 512
     messages:
-      - content: You are a knowledgable, engaging, history teacher.
+      - content: You are a knowledgable, engaging, biology teacher.
         role: system
-      - content: What was the role of falcons in Lord of the Rings?
+      - content: What makes falcons effective hunters?
         role: user
     stream: true
-    temperature: 0.3
-    top_k: 40
-    top_p: 1.0
+    temperature: 0.6
   repo_id: tiiuae/Falcon3-10B-Instruct
-model_name: Falcon3-10B-Instruct-with-1B-SpecDec
+model_name: Falcon 3 10B Instruct
 resources:
   accelerator: H100
   cpu: "1"

falcon/falcon3-3B-trt-llm-engine-high-throughput/config.yaml

@@ -2,26 +2,22 @@ build_commands: []
 environment_variables: {}
 external_package_dirs: []
 model_metadata:
-  tags:
-    - openai-compatible
   example_model_input:
     frequency_penalty: 1
     max_tokens: 512
     messages:
-      - content: You are a knowledgable, engaging, history teacher.
+      - content: You are a knowledgable, engaging, biology teacher.
         role: system
-      - content: What was the role of falcons in Lord of the Rings?
+      - content: What makes falcons effective hunters?
         role: user
     stream: true
-    temperature: 0.3
-    top_k: 40
-    top_p: 1.0
+    temperature: 0.6
   repo_id: tiiuae/Falcon3-3B-Instruct
-model_name: Falcon3-3B-Instruct-H100
+model_name: Falcon 3 3B Instruct
 python_version: py39
 requirements: []
 resources:
-  accelerator: H100
+  accelerator: A10G
   cpu: "1"
   memory: 24Gi
   use_gpu: true
@@ -43,4 +39,4 @@ trt_llm:
     batch_scheduler_policy: max_utilization
     enable_chunked_context: true
     kv_cache_free_gpu_mem_fraction: 0.85
-    request_default_max_tokens: 131072
+    request_default_max_tokens: 8192