初始化项目，由ModelHub XC社区提供模型

Model: Mathieu-Thomas-JOSSET/joke-finetome-model-gguf-phi4-20260112-081758 Source: Original Platform
2026-04-11 12:30:59 +08:00
commit 2c2240df42
13 changed files with 4725 additions and 0 deletions
--- a/inference/endpoint_recipe.json
+++ b/inference/endpoint_recipe.json
@@ -0,0 +1,18 @@
+{
+  "engine": "llama.cpp",
+  "recommended_endpoint_settings": {
+    "max_tokens_per_request": 1024,
+    "max_concurrent_requests": 2,
+    "notes": "Memory scales roughly with (max_concurrent_requests * max_tokens_per_request)."
+  },
+  "recommended_generation_defaults": {
+    "temperature": 1.2,
+    "top_p": 0.95,
+    "min_p": 0.05,
+    "repeat_penalty": 1.08,
+    "max_tokens": 2560
+  },
+  "chat_template": "phi4",
+  "gguf_file": "",
+  "gguf_quant": "q8_0"
+}