Studio: do not show Run for embedding-only non-GGUF models in the Model Hub (#7245)

* Studio: do not show Run for embedding-only non-GGUF models in the Model Hub

A downloaded embedding-only repo (sentence-transformers, feature-extraction)
reports canChat by safetensors format and classifies as supported, so the Model
Hub showed a Run button that dead-ends at load. Keep embedding-only non-GGUF
models out of the Run gate. GGUF is unaffected (llama.cpp resolves embed vs
generate at load time), and these models stay trainable.

* Gate embedding-only models on the pipeline tag, not just capabilities

An embedding repo whose name or tags also imply code, vision, audio or
reasoning (e.g. jina-embeddings-v2-base-code picks up code from its
-code suffix) slipped the embedding-only Run gate and dead-ended on a
chat load. Treat a feature-extraction or sentence-similarity pipeline
tag as authoritative for the gate; the change only ever widens it.

* studio: tighten comments in the hub embedding run guard

* Tighten comments in the hub embedding run guard

---------

Co-authored-by: danielhanchen <unslothshared@gmail.com>

studio/frontend/src/features/hub/catalog/model-inspector.tsx

@@ -59,6 +59,13 @@ import { ModelReadme } from "./model-readme";
 import { OwnerAvatar } from "./owner-avatar";
 import { AccessChip, CapabilityPill } from "./shared";
 
+// HF pipeline_tag values authoritative for embedding-only repos; capability
+// labels (code/vision/audio) can leak onto them via name or tags.
+const EMBEDDING_PIPELINE_TAGS: ReadonlySet<string> = new Set([
+  "feature-extraction",
+  "sentence-similarity",
+]);
+
 function ViewRepositoryButton({
   repoId,
   isDataset,
@@ -531,11 +538,27 @@ export const ModelInspector = memo(function ModelInspector({
     ? formatCompact(model.totalParams)
     : "N/A";
   const unslothSupported = unslothSupport.status !== "unsupported";
+  // Embedding-only non-GGUF repos have no generative head, so keep them out of
+  // the Run gate. Prefer the pipeline tag, else the capability heuristic.
+  const isEmbeddingOnly =
+    !model.isGguf &&
+    model.capabilities.some((c) => c.key === "embedding") &&
+    (EMBEDDING_PIPELINE_TAGS.has(model.pipelineTag?.toLowerCase() ?? "") ||
+      !model.capabilities.some(
+        (c) =>
+          c.key === "conversational" ||
+          c.key === "tools" ||
+          c.key === "reasoning" ||
+          c.key === "code" ||
+          c.key === "vision" ||
+          c.key === "audio",
+      ));
   // Chat-only hosts (no supported GPU / usable MLX) run inference only through
   // llama.cpp, so only GGUF is loadable.
   const canRunModel =
     !isDataset &&
     (model.runtimeCapabilities?.canChat ?? true) &&
+    !isEmbeddingOnly &&
     (model.isGguf || (!chatOnly && unslothSupported));
   const canTrainModel =
     !isDataset &&