pi-llama-cpp 0.3.3 → 0.3.4

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pi-llama-cpp",
-  "version": "0.3.3",
+  "version": "0.3.4",
   "description": "Pi extension for llama.cpp integration. Supports both router and single modes.",
   "keywords": [
     "pi",

package/src/models/baseModel.ts

@@ -7,7 +7,7 @@ import {
 } from "../constants";
 import { Mode } from "../enums/mode";
 import { Status } from "../enums/status";
-import { DataProperty, ModelsEndpoint } from "../interfaces/endpoints/models";
+import { DataProperty } from "../interfaces/endpoints/models";
 import { PropsEndpoint } from "../interfaces/endpoints/props";
 import { rpc } from "../tools/retriever";
 
@@ -97,11 +97,11 @@ export abstract class BaseModel {
    */
   async getContextSize(): Promise<number> {
     try {
-      const { data } = await rpc<ModelsEndpoint>(`/models`);
-      const model = data.find((d) => d.id === this.id);
-
-      const response = model?.meta?.n_ctx;
-      return response ?? DEFAULT_CTX;
+      const { default_generation_settings } = await rpc<PropsEndpoint>(
+        `/props?model=${this.id}`,
+      );
+      const { n_ctx } = default_generation_settings;
+      return n_ctx;
     } catch {
       return DEFAULT_CTX;
     }

package/src/models/singleModel.ts

@@ -1,23 +1,8 @@
-import { DEFAULT_CTX } from "../constants";
 import { Mode } from "../enums/mode";
-import { PropsEndpoint } from "../interfaces/endpoints/props";
-import { rpc } from "../tools/retriever";
 import { BaseModel } from "./baseModel";
 
 export class SingleModel extends BaseModel {
   get mode(): Mode {
     return Mode.SINGLE;
   }
-
-  async getContextSize(): Promise<number> {
-    try {
-      const { default_generation_settings } = await rpc<PropsEndpoint>(
-        `/props?model=${this.id}`,
-      );
-      const { n_ctx } = default_generation_settings;
-      return n_ctx;
-    } catch {
-      return DEFAULT_CTX;
-    }
-  }
 }

package/tests/routerModel.test.ts

@@ -130,14 +130,9 @@ describe("RouterModel context size extraction", () => {
         },
       ],
     });
-    // Second call: super.getContextSize() -> /models with meta.n_ctx
+    // Second call: super.getContextSize() -> /props?model=test-model with default_generation_settings.n_ctx
     mockRpc.mockResolvedValueOnce({
-      data: [
-        {
-          id: "test-model",
-          meta: { n_ctx: 4096 },
-        },
-      ],
+      default_generation_settings: { n_ctx: 4096 },
     });
 
     const model = new RouterModel(

package/tests/singleModel.test.ts

@@ -85,28 +85,19 @@ describe("SingleModel getStatus", () => {
 });
 
 describe("SingleModel getContextSize", () => {
-  it("should return n_ctx from /models endpoint meta", async () => {
+  it("should return n_ctx from /props endpoint default_generation_settings", async () => {
     mockRpc.mockResolvedValueOnce({
-      data: [{ id: "test", meta: { n_ctx: 8192 } }],
+      default_generation_settings: { n_ctx: 8192 },
     });
 
     const model = createModel();
     const ctxSize = await model.getContextSize();
 
     expect(ctxSize).toBe(8192);
-    expect(mockRpc).toHaveBeenCalledWith("/models");
-  });
-
-  it("should return DEFAULT_CTX when model not found in /models", async () => {
-    mockRpc.mockResolvedValueOnce({ data: [] });
-
-    const model = createModel();
-    const ctxSize = await model.getContextSize();
-
-    expect(ctxSize).toBe(DEFAULT_CTX);
+    expect(mockRpc).toHaveBeenCalledWith("/props?model=test");
   });
 
-  it("should return DEFAULT_CTX when /models fails", async () => {
+  it("should return DEFAULT_CTX when /props fails", async () => {
     mockRpc.mockRejectedValueOnce(new Error("Connection refused"));
 
     const model = createModel();