truss deployment example

truss/config.yaml

@@ -1,4 +1,7 @@
+# yaml-language-server: $schema=https://raw.githubusercontent.com/basetenlabs/truss/main/truss/config.schema.json
+
 model_name: glm-5.2-vision
+description: GLM-5.2-Vision FP8 with 1M context on 8 B200 GPUs.
 model_metadata:
   example_model_input:
     model: glm-5.2-vision
@@ -18,13 +21,12 @@ model_metadata:
 
 # Deploy GLM-5.2-Vision on Baseten with Truss.
 #
-# Unlike the internal deployments this was derived from, there is NO checkpoint
-# assembly here: the model_cache repo below is already the fully assembled
-# checkpoint (GLM-5.2 text + MoonViT vision tower + trained projector + processor
-# remote code), so start_server.sh only installs the plugin and launches SGLang.
+#   uvx truss push --config config.yaml
 #
-# To deploy:  truss push --remote <your-remote>
-#             truss push --config config_nvfp4.yaml --remote <your-remote>   # NVFP4 build
+# There is no checkpoint assembly here: the public Hugging Face repo below is
+# already the fully assembled checkpoint (GLM-5.2 text + MoonViT vision tower +
+# trained projector + processor remote code), so start_server.sh only installs
+# the plugin and launches SGLang.
 base_image:
   # SGLang 0.5.13, CUDA 13, Blackwell (sm_100). Ships the DSA sparse-attention and
   # trtllm-gen fp4-MoE kernels this checkpoint needs.
@@ -39,8 +41,7 @@ docker_server:
   server_port: 8000
 
 model_cache:
-  # The whole model, one repo — no assembly at serve time. For the smaller
-  # Blackwell-only NVFP4 build, push config_nvfp4.yaml instead.
+  # Public Hugging Face repo: no Hugging Face token or Baseten secret is needed.
   - repo_id: baseten/GLM-5.2-Vision-FP8
     revision: main
     volume_folder: glm5v
@@ -50,9 +51,6 @@ environment_variables:
   # Must match volume_folder above.
   GLM5V_CKPT: /app/model_cache/glm5v
 
-secrets:
-  hf_access_token: null
-
 resources:
   accelerator: B200:8
   use_gpu: true