add support for qwen2.5 coder 1.5b and 7b

abca3bfa3769a3e5e095dfb58660a71dc6254465 · 2024-09-30 18:42:25 +0400 · Alex Cheema

Files touched

M exo/models.py
M tinychat/examples/tinychat/index.html

Diff

commit abca3bfa3769a3e5e095dfb58660a71dc6254465
Author: Alex Cheema <alexcheema123@gmail.com>
Date:   Mon Sep 30 18:42:25 2024 +0400

    add support for qwen2.5 coder 1.5b and 7b
---
 exo/models.py                         | 6 ++++++
 tinychat/examples/tinychat/index.html | 2 ++
 2 files changed, 8 insertions(+)

diff --git a/exo/models.py b/exo/models.py
index 7bf424ce..1eaabed3 100644
--- a/exo/models.py
+++ b/exo/models.py
@@ -38,6 +38,12 @@ model_base_shards = {
   ### llava
   "llava-1.5-7b-hf": {"MLXDynamicShardInferenceEngine": Shard(model_id="llava-hf/llava-1.5-7b-hf", start_layer=0, end_layer=0, n_layers=32),},
   ### qwen
+  "qwen-2.5-coder-1.5b": {
+    "MLXDynamicShardInferenceEngine": Shard(model_id="mlx-community/Qwen2.5-Coder-1.5B-Instruct-4bit", start_layer=0, end_layer=0, n_layers=28),
+  },
+  "qwen-2.5-coder-7b": {
+    "MLXDynamicShardInferenceEngine": Shard(model_id="mlx-community/Qwen2.5-Coder-7B-Instruct-4bit", start_layer=0, end_layer=0, n_layers=28),
+  },
   "qwen-2.5-7b": {
     "MLXDynamicShardInferenceEngine": Shard(model_id="mlx-community/Qwen2.5-7B-Instruct-4bit", start_layer=0, end_layer=0, n_layers=28),
   },
diff --git a/tinychat/examples/tinychat/index.html b/tinychat/examples/tinychat/index.html
index 714bf339..9a5e07c2 100644
--- a/tinychat/examples/tinychat/index.html
+++ b/tinychat/examples/tinychat/index.html
@@ -43,6 +43,8 @@
 <option value="deepseek-coder-v2-lite">Deepseek Coder V2 Lite</option>
 <option value="deepseek-coder-v2.5">Deepseek Coder V2.5</option>
 <option value="llava-1.5-7b-hf">LLaVa 1.5 7B (Vision Model)</option>
+<option value="qwen-2.5-coder-1.5b">Qwen 2.5 Coder 1.5B</option>
+<option value="qwen-2.5-coder-7b">Qwen 2.5 Coder 7B</option>
 <option value="qwen-2.5-7b">Qwen 2.5 7B</option>
 <option value="qwen-2.5-math-7b">Qwen 2.5 7B (Math)</option>
 <option value="qwen-2.5-14b">Qwen 2.5 14B</option>

← 3e13e5ed upgrade mlx to 0.18.0 · back to Exo · clearer example docs and add 405b example 67f789b6 →