Improve model card: add paper badge, paper link, and sample usage

#7
by nielsr HF Staff - opened
Files changed (1) hide show
  1. README.md +54 -4
README.md CHANGED
@@ -1,11 +1,11 @@
1
  ---
2
- license: mit
3
  language:
4
  - en
 
 
5
  pipeline_tag: image-text-to-text
6
  tags:
7
  - multimodal
8
- library_name: transformers
9
  ---
10
 
11
  # Fara-7B: An Efficient Agentic Model for Computer Use
@@ -14,9 +14,11 @@ library_name: transformers
14
  [![Hugging Face Dataset](https://img.shields.io/badge/🤗-Dataset-yellow)](https://huggingface.co/datasets/microsoft/WebTailBench)
15
  [![Foundry](https://img.shields.io/badge/Azure-Foundry-0089D6)](https://aka.ms/foundry-fara-7b)
16
  [![Github](https://img.shields.io/badge/Github-181717?logo=github&logoColor=white)](https://github.com/microsoft/fara)
 
17
 
18
  [Official Microsoft Blog](https://www.microsoft.com/en-us/research/?p=1155843&preview=1&_ppp=0a22f3e916)<br>
19
  [Technical Report](https://aka.ms/fara-techreport)<br>
 
20
  [Github](https://github.com/microsoft/fara)<br>
21
  [Microsoft Foundry](https://ai.azure.com/explore/models/Fara-7B/version/1/registry/azureml-msr?tid=72f988bf-86f1-41af-91ab-2d7cd011db47)<br>
22
 
@@ -90,6 +92,33 @@ Critical points where the agent stops include entering personal info, completing
90
 
91
  ## 2. Usage
92
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
93
  ### 2.1 Primary Use Cases
94
 
95
  - Automating web tasks such as shopping, booking travel, restaurant reservations, info-seeking, or account workflows.
@@ -148,11 +177,32 @@ You are provided with function signatures within XML tags:
148
  "type": "function",
149
  "function": {
150
  "name": "computer_use",
151
- "description": "Use a mouse and keyboard to interact with a computer, and take screenshots.\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\n* The screen's resolution is 1428x896.\n* Whenever you intend to move the cursor to click on an element like an icon, you should consult a screenshot to determine the coordinates of the element before moving the cursor.\n* If you tried clicking on a program or link but it failed to load, even after waiting, try adjusting your cursor position so that the tip of the cursor visually falls on the element that you want to click.\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\n* When a separate scrollable container prominently overlays the webpage, if you want to scroll within it, you typically need to mouse_move() over it first and then scroll().\n* If a popup window appears that you want to close, if left_click() on the 'X' or close button doesn't work, try key(keys=['Escape']) to close it.\n* On some search bars, when you type(), you may need to press_enter=False and instead separately call left_click() on the search button to submit the search query. This is especially true of search bars that have auto-suggest popups for e.g. locations\n* For calendar widgets, you usually need to left_click() on arrows to move between months and left_click() on dates to select them; type() is not typically used to input dates there.",
 
 
 
 
 
 
 
 
 
 
152
  "parameters": {
153
  "properties": {
154
  "action": {
155
- "description": "The action to perform. The available actions are:\n* key: Performs key down presses on the arguments passed in order, then performs key releases in reverse order. Includes 'Enter', 'Alt', 'Shift', 'Tab', 'Control', 'Backspace', 'Delete', 'Escape', 'ArrowUp', 'ArrowDown', 'ArrowLeft', 'ArrowRight', 'PageDown', 'PageUp', 'Shift', etc.\n* type: Type a string of text on the keyboard.\n* mouse_move: Move the cursor to a specified (x, y) pixel coordinate on the screen.\n* left_click: Click the left mouse button.\n* scroll: Performs a scroll of the mouse scroll wheel.\n* visit_url: Visit a specified URL.\n* web_search: Perform a web search with a specified query.\n* history_back: Go back to the previous page in the browser history.\n* pause_and_memorize_fact: Pause and memorize a fact for future reference.\n* wait: Wait specified seconds for the change to happen.\n* terminate: Terminate the current task and report its completion status.",
 
 
 
 
 
 
 
 
 
 
 
156
  "enum": ["key", "type", "mouse_move", "left_click", "scroll", "visit_url", "web_search", "history_back", "pause_and_memorize_fact", "wait", "terminate"],
157
  "type": "string"
158
  },
 
1
  ---
 
2
  language:
3
  - en
4
+ library_name: transformers
5
+ license: mit
6
  pipeline_tag: image-text-to-text
7
  tags:
8
  - multimodal
 
9
  ---
10
 
11
  # Fara-7B: An Efficient Agentic Model for Computer Use
 
14
  [![Hugging Face Dataset](https://img.shields.io/badge/🤗-Dataset-yellow)](https://huggingface.co/datasets/microsoft/WebTailBench)
15
  [![Foundry](https://img.shields.io/badge/Azure-Foundry-0089D6)](https://aka.ms/foundry-fara-7b)
16
  [![Github](https://img.shields.io/badge/Github-181717?logo=github&logoColor=white)](https://github.com/microsoft/fara)
17
+ [![Paper](https://img.shields.io/badge/Paper-2511.19663-red)](https://huggingface.co/papers/2511.19663)
18
 
19
  [Official Microsoft Blog](https://www.microsoft.com/en-us/research/?p=1155843&preview=1&_ppp=0a22f3e916)<br>
20
  [Technical Report](https://aka.ms/fara-techreport)<br>
21
+ [Paper](https://huggingface.co/papers/2511.19663)<br>
22
  [Github](https://github.com/microsoft/fara)<br>
23
  [Microsoft Foundry](https://ai.azure.com/explore/models/Fara-7B/version/1/registry/azureml-msr?tid=72f988bf-86f1-41af-91ab-2d7cd011db47)<br>
24
 
 
92
 
93
  ## 2. Usage
94
 
95
+ ### Sample Usage
96
+
97
+ You can try Fara-7B locally by setting up the environment and hosting the model. For full instructions, refer to the [GitHub repository](https://github.com/microsoft/fara#installation).
98
+
99
+ ```bash
100
+ # 1. Clone repository
101
+ git clone https://github.com/microsoft/fara.git
102
+ cd fara
103
+
104
+ # 2. Setup environment
105
+ python3 -m venv .venv
106
+ source .venv/bin/activate
107
+ pip install -e .
108
+ playwright install
109
+ ```
110
+
111
+ Then in one process, host the model:
112
+ ```bash
113
+ vllm serve "microsoft/Fara-7B" --port 5000 --dtype auto
114
+ ```
115
+ Then you can iterative query it with:
116
+ ```bash
117
+ fara-cli --task "whats the weather in new york now"
118
+ ```
119
+
120
+ Hint: might need to do `--tensor-parallel-size 2` with vllm command if you run out of memory
121
+
122
  ### 2.1 Primary Use Cases
123
 
124
  - Automating web tasks such as shopping, booking travel, restaurant reservations, info-seeking, or account workflows.
 
177
  "type": "function",
178
  "function": {
179
  "name": "computer_use",
180
+ "description": "Use a mouse and keyboard to interact with a computer, and take screenshots.\
181
+ * This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\
182
+ * Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\
183
+ * The screen's resolution is 1428x896.\
184
+ * Whenever you intend to move the cursor to click on an element like an icon, you should consult a screenshot to determine the coordinates of the element before moving the cursor.\
185
+ * If you tried clicking on a program or link but it failed to load, even after waiting, try adjusting your cursor position so that the tip of the cursor visually falls on the element that you want to click.\
186
+ * Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\
187
+ * When a separate scrollable container prominently overlays the webpage, if you want to scroll within it, you typically need to mouse_move() over it first and then scroll().\
188
+ * If a popup window appears that you want to close, if left_click() on the 'X' or close button doesn't work, try key(keys=['Escape']) to close it.\
189
+ * On some search bars, when you type(), you may need to press_enter=False and instead separately call left_click() on the search button to submit the search query. This is especially true of search bars that have auto-suggest popups for e.g. locations\
190
+ * For calendar widgets, you usually need to left_click() on arrows to move between months and left_click() on dates to select them; type() is not typically used to input dates there.",
191
  "parameters": {
192
  "properties": {
193
  "action": {
194
+ "description": "The action to perform. The available actions are:\
195
+ * key: Performs key down presses on the arguments passed in order, then performs key releases in reverse order. Includes 'Enter', 'Alt', 'Shift', 'Tab', 'Control', 'Backspace', 'Delete', 'Escape', 'ArrowUp', 'ArrowDown', 'ArrowLeft', 'ArrowRight', 'PageDown', 'PageUp', 'Shift', etc.\
196
+ * type: Type a string of text on the keyboard.\
197
+ * mouse_move: Move the cursor to a specified (x, y) pixel coordinate on the screen.\
198
+ * left_click: Click the left mouse button.\
199
+ * scroll: Performs a scroll of the mouse scroll wheel.\
200
+ * visit_url: Visit a specified URL.\
201
+ * web_search: Perform a web search with a specified query.\
202
+ * history_back: Go back to the previous page in the browser history.\
203
+ * pause_and_memorize_fact: Pause and memorize a fact for future reference.\
204
+ * wait: Wait specified seconds for the change to happen.\
205
+ * terminate: Terminate the current task and report its completion status.",
206
  "enum": ["key", "type", "mouse_move", "left_click", "scroll", "visit_url", "web_search", "history_back", "pause_and_memorize_fact", "wait", "terminate"],
207
  "type": "string"
208
  },