scholarly360
commited on
Commit
•
09fea05
1
Parent(s):
818925c
Update README.md
Browse files
README.md
CHANGED
@@ -15,13 +15,16 @@ This modelcard aims to be a base template for new models. It has been generated
|
|
15 |
|
16 |
## Model Details
|
17 |
|
|
|
|
|
18 |
### Model Description
|
19 |
|
20 |
<!-- Provide a longer summary of what this model is. -->
|
21 |
|
22 |
-
This model is fine-tuned using Alpaca like instructions. The base data for instruction fine-tuning
|
23 |
|
24 |
-
|
|
|
25 |
|
26 |
|
27 |
- **Developed by:** [More Information Needed]
|
@@ -42,6 +45,7 @@ An encoder-decoder architecture like flag T5 is used because the author found it
|
|
42 |
## Uses
|
43 |
|
44 |
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
|
|
45 |
|
46 |
### Direct Use
|
47 |
|
@@ -82,17 +86,24 @@ Use the code below to get started with the model.
|
|
82 |
>>> model_name = "scholarly360/contracts-extraction-flan-t5-base"
|
83 |
>>> model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
|
84 |
>>> tokenizer = AutoTokenizer.from_pretrained(model_name)
|
85 |
-
>>>
|
86 |
-
>>> prompt = """ what kind of clause is "Neither Party shall be liable to the other for any abatement of Charges, delay or non-performance of its obligations under the Services Agreement arising from any cause or causes beyond its reasonable control (a
|
87 |
>>> inputs = tokenizer(prompt, return_tensors="pt")
|
88 |
>>> outputs = model.generate(**inputs)
|
89 |
>>> print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
|
90 |
-
>>>
|
91 |
-
>>> prompt = """ what is agreement date in
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
92 |
>>> inputs = tokenizer(prompt, return_tensors="pt")
|
93 |
>>> outputs = model.generate(**inputs)
|
94 |
>>> print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
|
95 |
-
|
96 |
```
|
97 |
|
98 |
|
@@ -103,7 +114,7 @@ Use the code below to get started with the model.
|
|
103 |
### Training Data
|
104 |
|
105 |
<!-- This should link to a Data Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
106 |
-
|
107 |
[More Information Needed]
|
108 |
|
109 |
### Training Procedure
|
|
|
15 |
|
16 |
## Model Details
|
17 |
|
18 |
+
Instruction fine tuned Flan-T5 on Contracts
|
19 |
+
|
20 |
### Model Description
|
21 |
|
22 |
<!-- Provide a longer summary of what this model is. -->
|
23 |
|
24 |
+
This model is fine-tuned using Alpaca like instructions. The base data for instruction fine-tuning is a legal corpus with fields like Titles , agreement date, party name, and addresses.
|
25 |
|
26 |
+
There are many type of models trained on above DataSet (DataSet will be released soon for the community)
|
27 |
+
An encoder-decoder architecture like Flan-T5 is used because the author found it to be better than a decoder only architecture given the same number of parameters.
|
28 |
|
29 |
|
30 |
- **Developed by:** [More Information Needed]
|
|
|
45 |
## Uses
|
46 |
|
47 |
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
48 |
+
Just like any ChatGPT equivalent model (For Contracts Domain)
|
49 |
|
50 |
### Direct Use
|
51 |
|
|
|
86 |
>>> model_name = "scholarly360/contracts-extraction-flan-t5-base"
|
87 |
>>> model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
|
88 |
>>> tokenizer = AutoTokenizer.from_pretrained(model_name)
|
89 |
+
>>> ### Example 1
|
90 |
+
>>> prompt = """ what kind of clause is "Neither Party shall be liable to the other for any abatement of Charges, delay or non-performance of its obligations under the Services Agreement arising from any cause or causes beyond its reasonable control (a Force Majeure Event) including, without limitation """
|
91 |
>>> inputs = tokenizer(prompt, return_tensors="pt")
|
92 |
>>> outputs = model.generate(**inputs)
|
93 |
>>> print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
|
94 |
+
>>> ### Example 1
|
95 |
+
>>> prompt = """ what is agreement date in 'This COLLABORATION AGREEMENT (Agreement) dated November 14, 2002, is made by and between ZZZ, INC., a Delaware corporation' """"
|
96 |
+
>>> inputs = tokenizer(prompt, return_tensors="pt")
|
97 |
+
>>> outputs = model.generate(**inputs)
|
98 |
+
>>> print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
|
99 |
+
>>> ### Example 3
|
100 |
+
>>> prompt = """ ### Instruction:
|
101 |
+
what is agreement date
|
102 |
+
### Input:
|
103 |
+
This COLLABORATION AGREEMENT (Agreement) dated November 14, 2002, is made by and between ZZZ, INC., a Delaware corporation """"
|
104 |
>>> inputs = tokenizer(prompt, return_tensors="pt")
|
105 |
>>> outputs = model.generate(**inputs)
|
106 |
>>> print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
|
|
|
107 |
```
|
108 |
|
109 |
|
|
|
114 |
### Training Data
|
115 |
|
116 |
<!-- This should link to a Data Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
117 |
+
DataSet will be released soon for the community
|
118 |
[More Information Needed]
|
119 |
|
120 |
### Training Procedure
|