Switch to MPT

.gitignore

@@ -3,6 +3,7 @@
 
 # Environment
 .envrc
+.env
 
 # Editor
 pyrightconfig.json

deployment/docker-compose-0.yaml

@@ -1,8 +1,8 @@
 services:
-  Falcon-7B:
+  MPT-7B:
     container_name: worker0
     image: mlenergy/tgi:latest
-    command: ["--model-id", "tiiuae/falcon-7b-instruct", "--num-shard", "1", "--otlp-endpoint", "http://jaeger:4317"]
+    command: ["--model-id", "mosaicml/mpt-7b-chat", "--num-shard", "1", "--otlp-endpoint", "http://jaeger:4317"]
     shm_size: 1g
     networks:
       - leaderboard

deployment/docker-compose-1.yaml

@@ -5,7 +5,7 @@ services:
     command: ["--model-id", "meta-llama/Llama-2-70b-chat-hf", "--num-shard", "2", "--otlp-endpoint", "http://jaeger:4317", "--quantize", "bitsandbytes"]
     shm_size: 1g
     environment:
-      HUGGING_FACE_HUB_TOKEN: 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
+      HUGGING_FACE_HUB_TOKEN: "${HF_TOKEN}"
     networks:
       - leaderboard
     volumes:
@@ -19,10 +19,10 @@ services:
             - driver: nvidia
               device_ids: ["0", "1"]
               capabilities: [gpu]
-  Falcon-40B:
+  MPT-30B:
     container_name: worker5
     image: mlenergy/tgi:latest
-    command: ["--model-id", "tiiuae/falcon-40b-instruct", "--num-shard", "2", "--otlp-endpoint", "http://jaeger:4317"]
+    command: ["--model-id", "mosaicml/mpt-30b-chat", "--num-shard", "2", "--otlp-endpoint", "http://jaeger:4317"]
     shm_size: 1g
     networks:
       - leaderboard