Upload folder using huggingface_hub

README.md

@@ -9,5 +9,42 @@ tags:
 # phi-2-MNN
 
 ## Introduction
+This model is a 4-bit quantized version of the MNN model exported from [phi-2](https://modelscope.cn/models/mengzhao/phi-2/summary) using [llmexport](https://github.com/alibaba/MNN/tree/master/transformers/llm/export).
 
-This model is a 4-bit quantized version of the MNN model exported from phi-2 using [llm-export](https://github.com/wangzhaode/llm-export).
+## Download
+```bash
+# install huggingface
+pip install huggingface
+```
+```bash
+# shell download
+huggingface download --model 'taobao-mnn/phi-2-MNN' --local_dir 'path/to/dir'
+```
+```python
+# SDK download
+from huggingface_hub import snapshot_download
+model_dir = snapshot_download('taobao-mnn/phi-2-MNN')
+```
+
+```bash
+# git clone
+git clone https://www.modelscope.cn/taobao-mnn/phi-2-MNN
+```
+
+## Usage
+```bash
+# clone MNN source
+git clone https://github.com/alibaba/MNN.git
+
+# compile
+cd MNN
+mkdir build && cd build
+cmake .. -DMNN_LOW_MEMORY=true -DMNN_CPU_WEIGHT_DEQUANT_GEMM=true -DMNN_BUILD_LLM=true -DMNN_SUPPORT_TRANSFORMER_FUSE=true
+make -j
+
+# run
+./llm_demo /path/to/phi-2-MNN/config.json prompt.txt
+```
+
+## Document
+[MNN-LLM](https://mnn-docs.readthedocs.io/en/latest/transformers/llm.html#)

config.json

@@ -1,9 +1,8 @@
 {
-    "llm_model": "phi2-int4.mnn",
-    "llm_weight": "phi2-int4.mnn.weight",
-
+    "llm_model": "llm.mnn",
+    "llm_weight": "llm.mnn.weight",
     "backend_type": "cpu",
     "thread_num": 4,
     "precision": "low",
     "memory": "low"
-}
+}

llm.mnn

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60f2064d6235755629ccb91177177dffa330c2dd64c9dc0ff3a00764abc58169
+size 1195984

llm.mnn.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90c13b5d8b59642a9eea79a4014d121e4e876b4d9bb295b495b451cbded3f44f
+size 7325129

llm.mnn.weight

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:809ed6b0670b761921efa79989ac850a9e1e0891b94c28b2013a5abc1cfd9311
+size 1492463766

llm_config.json

@@ -1,15 +1,14 @@
 {
     "hidden_size": 2560,
     "layer_nums": 32,
-    "attention_mask": "glm",
+    "attention_mask": "float",
     "key_value_shape": [
+        2,
         1,
         0,
-        2,
         32,
         80
     ],
     "prompt_template": "Instruct: %s\nOutput:",
-    "is_visual": false,
-    "is_single": true
+    "is_visual": false
 }