update license now that model has been updated

README.md

@@ -10,20 +10,49 @@ tags:
 - llama
 - llama-2
 - functions
+- function calling
 ---
-# Llama 2 - 7B Model With Function Calling Capabilities
+# fLlama 2 - Function Calling Llama 2 - 7B
 
 ## Description
 
-This model extends the Llama 2 - 7B model with function calling capabilities. It allows users to make function calls within the model's input, making it capable of performing a variety of tasks.
+This model extends the hugging face Llama 2 - 7B model with function calling capabilities. It allows users to make function calls within the model's input, making it capable of performing a variety of tasks.
 
-!!! This is an early proof of concept. Please post issues and questions !!!
+## Licensing and Usage
 
-Use of this model is governed by the [Meta license](https://ai.meta.com/resources/models-and-libraries/llama-downloads/).
+### Non-commercial Use
+Trelis' **fLlama 2** is licensed for non-commercial use Creative Commons Attribution-NonCommercial (CC BY-NC)*.
 
-## Usage
+Some examples of non-commercial use:
+- Research within an academic institution
+- Personal research
 
-Firstly, ensure you have the `transformers` package installed:
+### Commercial Use
+You can [purchase a life-time license here](https://buy.stripe.com/7sI14QcWhci71ck3ce) for commercial use for €9.99 per GPU/TPU. Commercial use of fLlama on a CPU is allowed for free.
+
+*In addition to the above, all use of this model is further subject to terms in the [Meta license](https://ai.meta.com/resources/models-and-libraries/llama-downloads/).
+
+Some examples of commercial use:
+- Including or building on fLlama within a product or service, whether monetized or not.
+- Using or building on fLlama for an internal company language model.
+
+Commercial license examples:
+- Running fLlama on a CPU = free
+- Running fLlama on a CPU + GPU/TPU = 1 license required.
+- Running fLlama sharded across 2 GPUs = 2 licenses required.
+Licenses are transferable across different machines provided they are not running simultaneously.
+
+Model weights are provided as-is, without warranty, and with the same disclaimers as provided by Meta (see Llama 2 below).
+
+### Dataset Licensing
+
+The dataset used for training this model can be found at [Trelis Function Calling Dataset] (https://huggingface.co/datasets/Trelis/function_calling).
+
+The data may be used for free for commercial and non-commercial use. It is Apache 2 licensed and does not have any dependencies (i.e. does not depend on Meta's Llama license.)
+
+## Inference with HuggingFace 🤗
+
+Ensure you have the `transformers` package installed:
 
 ```
 pip install transformers
@@ -41,58 +70,79 @@ model = AutoModelForCausalLM.from_pretrained("Trelis/Llama-2-7b-chat-hf-function
 To make a function call, you should format your input like this:
 
 ```
-input_text = f'<<SYS>>You are a helpful research assistant. The following functions, if any, are available for you to fetch further data to answer user questions: {function_metadata}<</SYS>>[INST]Your user command here[/INST]'
-```
+<s> [INST] <<SYS>>
+You are a helpful research assistant. The following functions are available for you to fetch further data to answer user questions, if relevant:
 
-Function metadata should be a string representation of a JSON object, like this:
+{
+    "function": "search_bing",
+    "description": "Search the web for content on Bing. This allows users to search online/the internet/the web for content.",
+    "arguments": [
+        {
+            "name": "query",
+            "type": "string",
+            "description": "The search query string"
+        }
+    ]
+}
 
-```
-function_metadata = ```
 {
-  "name": "search_bing",
-  "description": "Search the web for content on Bing. This allows users to search online/the internet/the web for content.",
-  "parameters": {
-    "type": "object",
-    "properties": {
-      "query": {
-      "type": "string",
-      "description": "The search query string"
-      }
-  },
-  "required": ["query"]
-  }
+    "function": "search_arxiv",
+    "description": "Search for research papers on ArXiv. Make use of AND, OR and NOT operators as appropriate to join terms within the query.",
+    "arguments": [
+        {
+            "name": "query",
+            "type": "string",
+            "description": "The search query string"
+        }
+    ]
 }
-'''
-```
+                
 
-An example command would be:
+To call a function, respond - immediately and only - with a JSON object of the following format:
+{
+    "function": "function_name",
+    "arguments": {
+        "argument1": "argument_value",
+        "argument2": "argument_value"
+    }
+}
+<</SYS>>
 
-```
-command = "Search the internet for Irish stew recipes."
+Find papers on high pressure batch reverse osmosis [/INST]
 ```
 
-So, the whole input would look like this:
+Notice that functionMetadata should be a string representation of a JSON object, like this:
 
 ```
-input_text = f'<<SYS>>You are a helpful research assistant. The following functions, if any, are available for you to fetch further data to answer user questions: {function_metadata}<</SYS>>[INST]{command}[/INST]'
+"functionMetaData": {
+        "function": "search_bing",
+        "description": "Search the web for content on Bing. This allows users to search online/the internet/the web for content.",
+        "arguments": [
+            {
+                "name": "query",
+                "type": "string",
+                "description": "The search query string"
+            }
+        ]
+    }
+'''
 ```
 
-The model's response will be a JSON object containing the function call. Here's how you can parse it:
-
+and the language model should respond with a json object formatted like this:
 ```
-response_json = json.loads(response)
+{
+    "function": "function_name",
+    "arguments": {
+        "argument1": "argument_value",
+        "argument2": "argument_value"
+    }
+}
 ```
 
-You can then use response_json to access the function call and arguments.
-
 It is recommended to handle cases where:
 - There is no json object in the response
 - The response contains text in addition to the json response
 
-## Dataset
-
-The dataset used for training this model can be found at [Trelis Function Calling Dataset] (https://huggingface.co/datasets/Trelis/function_calling).
-
 ## Quanitization Configurations
 
 The following `bitsandbytes` quantization config was used during training:
@@ -119,7 +169,7 @@ The following `bitsandbytes` quantization config was used during training:
 
 ~
 
-All details below are copied from the original repo.
+Below follows information on the original Llama 2 model...
 
 ~