jayr014 commited on
Commit
80a88bc
1 Parent(s): ad5e37c

Updating Readme

Browse files
Files changed (1) hide show
  1. README.md +124 -39
README.md CHANGED
@@ -4,11 +4,11 @@
4
  license: apache-2.0
5
  ---
6
 
7
- # BloomChat V1.0
8
 
9
  <!-- Provide a quick summary of what the model is/does. -->
10
 
11
- BloomChat-v1.0 is based on [BigScience Group Bloom-176 model](https://huggingface.co/bigscience/bloom), and is instruction-tuned on a subset of 100k datapoints per data source from the [OIG dataset](https://huggingface.co/datasets/laion/OIG) provided by laion. Then aligned using [Dolly 2.0](https://huggingface.co/datasets/databricks/databricks-dolly-15k) and [Oasst1](https://huggingface.co/datasets/OpenAssistant/oasst1).
12
 
13
  ## Model Details
14
 
@@ -18,15 +18,14 @@ BloomChat-v1.0 is based on [BigScience Group Bloom-176 model](https://huggingfac
18
 
19
  - **Developed by:** [SambaNova Systems](https://sambanova.ai/) and [Together Computer](https://www.together.xyz/)
20
  - **Model type:** Language Model
21
- - **Language(s):** Multiple; see [training data from Bloom-176B](https://huggingface.co/bigscience/bloom#training-data)
22
- - **License:** apache-2.0
23
- - **Instruction Tuned from model:** [BigScience Group Bloom-176B](https://huggingface.co/bigscience/bloom)
24
 
25
  ### Additional Information
26
 
27
  <!-- Provide the basic links for the model. -->
28
-
29
- - **Blogpost:** [More Information Needed]
30
 
31
  ## Uses
32
 
@@ -35,51 +34,40 @@ BloomChat-v1.0 is based on [BigScience Group Bloom-176 model](https://huggingfac
35
  ### Direct Use
36
 
37
  <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
 
38
 
39
- [More Information Needed]
40
-
41
- ### Downstream Use [optional]
42
-
43
- <!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
44
-
45
- [More Information Needed]
46
 
47
  ### Out-of-Scope Use
48
 
49
  <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
50
 
51
- [More Information Needed]
52
-
53
- ## Bias, Risks, and Limitations
54
-
55
- <!-- This section is meant to convey both technical and sociotechnical limitations. -->
56
-
57
- Like all LLMs, BloomChat has certain limitations:
58
- - Hallucination: BloomChat may sometimes generate responses that contain plausible-sounding but factually incorrect or irrelevant information.
59
- - Code Switching: The model might unintentionally switch between languages or dialects within a single response, affecting the coherence and understandability of the output.
60
- - Repetition: BloomChat may produce repetitive phrases or sentences, leading to less engaging and informative responses.
61
- - Coding and Math: The model's performance in generating accurate code or solving complex mathematical problems may be limited.
62
- - Toxicity: BloomChat may inadvertently generate responses containing inappropriate or harmful content.
63
 
64
  ### Recommendations
65
 
66
  <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
67
 
68
- Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model.
69
 
70
  ## How to Get Started with the Model
71
 
72
  Use the code below to get started with the model.
73
 
74
- [More Information Needed]
 
 
 
 
 
 
75
 
76
- ### Suggested inference parameters
77
  - Temperature: 0.8
78
  - Repetition penalty: 1.2
79
  - Top-p: 0.9
80
  - Max generated tokens: 512
81
 
82
- ### Suggested System Prompts
83
  ```
84
  <human>: Write a script in which Bob accidentally breaks his dad's guitar
85
  <bot>:
@@ -102,6 +90,75 @@ Context: U.S. authorities launched emergency measures on Sunday to shore up conf
102
  <bot>:
103
  ```
104
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
105
  ## Training Details
106
 
107
  ### Training Data
@@ -116,7 +173,7 @@ Context: U.S. authorities launched emergency measures on Sunday to shore up conf
116
 
117
  <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
118
 
119
- We trained BloomChat with SambaStudio, a platform built on SambaNova's in-house Reconfigurable Dataflow Unit (RDU). We started from [Bloom-176B](https://huggingface.co/bigscience/bloom), an OSS multilingual 176B GPT model pretrained by the [BigScience group](https://huggingface.co/bigscience).
120
 
121
  ### Prompting Style Used For Training
122
  ```
@@ -163,21 +220,49 @@ We trained BloomChat with SambaStudio, a platform built on SambaNova's in-house
163
  - End Learning Ratio: 0.1
164
  - Weight decay: 0.1
165
 
 
166
 
167
- ## Evaluation
168
 
169
- <!-- This section describes the evaluation protocols and provides the results. -->
 
 
 
 
 
 
 
 
 
 
 
 
 
170
 
171
- ![HELM core-scenarios](images/HELM_core-senarios_CNN+MS_Marco_WIP.png)
172
 
173
- ![Multilingual scores French and hindi](images/Multilinguality_WMT-14_on_French+Hindi.png)
174
 
175
- ![Multilingual scores Chinese](images/Multilinguality_WMT-14_on_Simplified_Chinese.png)
 
 
 
 
 
176
 
177
- ![Mean Win Rate on HELM](images/Open_source_model_Mean_Win_Rate_on_HELM_core_scenarios.png)
178
 
179
- ## Community
180
 
181
- [Link to discord server]
182
 
 
 
 
 
 
 
 
 
 
183
 
 
4
  license: apache-2.0
5
  ---
6
 
7
+ # BLOOMChat V1.0
8
 
9
  <!-- Provide a quick summary of what the model is/does. -->
10
 
11
+ BLOOMChat is based on [BigScience Group BLOOM model](https://huggingface.co/bigscience/bloom), and is instruction-tuned on a subset of 100k datapoints per data source from the [OIG dataset](https://huggingface.co/datasets/laion/OIG) from the [OpenChatKit](https://www.together.xyz/blog/openchatkit). Then aligned using [Dolly 2.0](https://huggingface.co/datasets/databricks/databricks-dolly-15k) and [Oasst1](https://huggingface.co/datasets/OpenAssistant/oasst1).
12
 
13
  ## Model Details
14
 
 
18
 
19
  - **Developed by:** [SambaNova Systems](https://sambanova.ai/) and [Together Computer](https://www.together.xyz/)
20
  - **Model type:** Language Model
21
+ - **Language(s):** Multiple; see [training data from BLOOM](https://huggingface.co/bigscience/bloom#training-data)
22
+ - **License:** apache-2.0 with RAIL restrictions
23
+ - **Instruction Tuned from model:** [BigScience Group BLOOM](https://huggingface.co/bigscience/bloom)
24
 
25
  ### Additional Information
26
 
27
  <!-- Provide the basic links for the model. -->
28
+ - **Blog Post**: [More Information Needed]
 
29
 
30
  ## Uses
31
 
 
34
  ### Direct Use
35
 
36
  <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
37
+ This model is intended for commercial and research use.
38
 
 
 
 
 
 
 
 
39
 
40
  ### Out-of-Scope Use
41
 
42
  <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
43
 
44
+ BLOOMChat is intended for chatbot applications and may not perform well in use cases outside of the intended use. BLOOMChat should NOT be used for safety-critical applications or for making decisions that have a significant impact on individuals or society.
 
 
 
 
 
 
 
 
 
 
 
45
 
46
  ### Recommendations
47
 
48
  <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
49
 
50
+ Users should be made aware of the risks, biases, limitations, and restrictions of the model, which are listed down at the bottom of the page.
51
 
52
  ## How to Get Started with the Model
53
 
54
  Use the code below to get started with the model.
55
 
56
+ ```python
57
+ from transformers import AutoModelForCausalLM, AutoTokenizer
58
+
59
+ tokenizer = AutoTokenizer.from_pretrained("bloomchat/BloomChat")
60
+ model = AutoModelForCausalLM.from_pretrained("bloomchat/BloomChat", device_map="auto", torch_dtype="auto")
61
+ ```
62
+
63
 
64
+ ### Suggested Inference Parameters
65
  - Temperature: 0.8
66
  - Repetition penalty: 1.2
67
  - Top-p: 0.9
68
  - Max generated tokens: 512
69
 
70
+ ### Suggested Prompts To Try
71
  ```
72
  <human>: Write a script in which Bob accidentally breaks his dad's guitar
73
  <bot>:
 
90
  <bot>:
91
  ```
92
 
93
+ ### Some example completions
94
+ ```
95
+ <human>: What does it mean for the Fed to raise or lower interest rates?
96
+ <bot>: The Federal Reserve (also known as the Fed) is the central banking system of the United States. It was created in 1913, and is an independent agency of the federal government. The Fed has a dual mandate of maximizing employment and stabilizing prices. The Fed sets interest rates, and can influence the money supply and the credit markets through its actions.
97
+
98
+ Raising interest rates means that the Fed is trying to reduce the money supply in the economy, and is trying to discourage borrowing and encourage saving. Lowering interest rates means that the Fed is trying to stimulate economic growth by encouraging borrowing and reducing the cost of credit.
99
+
100
+ The Fed raises or lowers interest rates in order to influence the economy and achieve its goals for employment and price stability. Interest rates are one of the primary tools that the Fed uses to achieve its goals.
101
+ ```
102
+
103
+ ```
104
+ <human>: Write an engaging and well-written property listing description for selling a house. Address of the house and some of the details are given to you. Fill in the information gap with hallucinations if needed.
105
+
106
+ Property Address: 412 Monterey Ave, Capitola, CA 95010
107
+ Parking: 3 cars + a finished single car garage
108
+ Details:
109
+ - open floorplan
110
+ - window with views to park/mountains
111
+ - Kitchen with shaker cabinetry and white Calcatta quartz counters
112
+ <bot>: This home at 412 Monterey Ave, Capitola, CA 95010 is the perfect place to call home. With its open floor plan, it provides the perfect space for entertaining guests. The kitchen is equipped with shaker cabinetry and white Calcatta quartz counters, providing the perfect space to create culinary masterpieces. The window with views to the park and mountains provide a beautiful view of the outdoors, making this home feel like a piece of paradise. The three-car garage and finished single car garage provide ample parking space, making this home a must-see for anyone looking for a new home.
113
+
114
+ ```
115
+
116
+ ```
117
+ <human>: Generate a table summarizing the options outlined in this email.
118
+
119
+ Team, we have to make a decision on the product plan for Goocrux v2.
120
+
121
+ The first option will take us 3 weeks of development, and will achieve performance of 80ms, but means that the UI has to change.
122
+
123
+ The second option will take us 4 weeks of development, and will achieve performance of 60ms, without any UI change.
124
+
125
+ The third option will take the longest -- 6 weeks of development, will achieve performance of 30ms, without any UI change.
126
+
127
+ # Table with four columns: (1) Option ID, (2) Development time, (3) Performance, (4) Require UI Change
128
+ <bot>: Option ID Development time Performance Require UI Change
129
+ Option 1 3 weeks 80ms Yes
130
+ Option 2 4 weeks 60ms No
131
+ Option 3 6 weeks 30ms No
132
+ ```
133
+ ```
134
+ <human>: Explain the difference between a proton and a neutron
135
+ <bot>: A proton and a neutron are both subatomic particles, but they differ in several ways:
136
+
137
+ Mass: A proton has a mass of approximately 1.67 x 10^-27 kilograms, while a neutron has a mass of approximately 1.67 x 10^-27 kilograms.
138
+
139
+ Charge: A proton has a positive charge, while a neutron has no charge.
140
+
141
+ Spinning: Protons and neutrons both have a spinning motion called intrinsic spin, but the direction of their spin is opposite to each other.
142
+
143
+ Protons are found in the nucleus of atoms, while neutrons are found in the nucleus of atoms.
144
+ In the nucleus of atoms, protons and neutrons are bound together in a structure called a nucleus, which gives the atom its stability.
145
+ ```
146
+ ## Evaluation
147
+
148
+ <!-- This section describes the evaluation protocols and provides the results. -->
149
+
150
+ ![Human evaluation](images/Human_evaluation.png)
151
+ <figure style="text-align:center;">
152
+ <figcaption><b>BLOOMChat vs Baselines Model in Human Preference Rankings</b></figcaption>
153
+ </figure>
154
+ <br>
155
+
156
+ ![Multilingual evaluation](images/Multilingual_capabilities_comparison.png)
157
+ <figure style="text-align:center;">
158
+ <figcaption><b>BLOOMChat surpasses other Bloom variants and state-of-the-art open chat models in cross-lingual setting</b></figcaption>
159
+ </figure>
160
+ <br>
161
+
162
  ## Training Details
163
 
164
  ### Training Data
 
173
 
174
  <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
175
 
176
+ We trained BLOOMChat with SambaStudio, a platform built on SambaNova's in-house Reconfigurable Dataflow Unit (RDU). We started from [BLOOM](https://huggingface.co/bigscience/bloom), an OSS multilingual 176B GPT model pretrained by the [BigScience group](https://huggingface.co/bigscience). There was also some preprocessing done on the training datasets.
177
 
178
  ### Prompting Style Used For Training
179
  ```
 
220
  - End Learning Ratio: 0.1
221
  - Weight decay: 0.1
222
 
223
+ ### RAIL Restrictions
224
 
225
+ As the original model [BLOOM](https://huggingface.co/bigscience/bloom) is on the RAIL License, we need to follow the same restrictions.
226
 
227
+ You agree not to use the Model or Derivatives of the Model:
228
+ 1. In any way that violates any applicable national, federal, state, local or international law or regulation;
229
+ 2. For the purpose of exploiting, harming or attempting to exploit or harm minors in any way;
230
+ 3. To generate or disseminate verifiably false information with the purpose of harming others;
231
+ 4. To generate or disseminate personal identifiable information that can be used to harm an individual;
232
+ 5. To generate or disseminate information or content, in any context (e.g. posts, articles, tweets, chatbots or other kinds of automated bots) without expressly and intelligibly disclaiming that the text is machine generated;
233
+ 6. To defame, disparage or otherwise harass others;
234
+ 7. To impersonate or attempt to impersonate others;
235
+ 8. For fully automated decision making that adversely impacts an individual’s legal rights or otherwise creates or modifies a binding, enforceable obligation;
236
+ 9. For any use intended to or which has the effect of discriminating against or harming individuals or groups based on online or offline social behavior or known or predicted personal or personality characteristics
237
+ 10. To exploit any of the vulnerabilities of a specific group of persons based on their age, social, physical or mental characteristics, in order to materially distort the behavior of a person pertaining to that group in a manner that causes or is likely to cause that person or another person physical or psychological harm;
238
+ 11. For any use intended to or which has the effect of discriminating against individuals or groups based on legally protected characteristics or categories;
239
+ 12. To provide medical advice and medical results interpretation;
240
+ 13. To generate or disseminate information for the purpose to be used for administration of justice, law enforcement, immigration or asylum processes, such as predicting an individual will commit fraud/crime commitment (e.g. by text profiling, drawing causal relationships between assertions made in documents, indiscriminate and arbitrarily-targeted use).
241
 
242
+ ### Bias, Risks, and Limitations
243
 
244
+ <!-- This section is meant to convey both technical and sociotechnical limitations. -->
245
 
246
+ Like all LLMs, BLOOMChat has certain limitations:
247
+ - Hallucination: BLOOMChat may sometimes generate responses that contain plausible-sounding but factually incorrect or irrelevant information.
248
+ - Code Switching: The model might unintentionally switch between languages or dialects within a single response, affecting the coherence and understandability of the output.
249
+ - Repetition: BLOOMChat may produce repetitive phrases or sentences, leading to less engaging and informative responses.
250
+ - Coding and Math: The model's performance in generating accurate code or solving complex mathematical problems may be limited.
251
+ - Toxicity: BLOOMChat may inadvertently generate responses containing inappropriate or harmful content.
252
 
253
+ ### Misuse and Malicious Use
254
 
255
+ BLOOMChat is designed for use in chatbot applications and should not be used for any other purpose. Misuse of the model, such as using it to engage in illegal or unethical activities, is strictly prohibited and goes against the principles of the BLOOMChat community project.
256
 
257
+ Using the model to generate content that is cruel to individuals is a misuse of this model. This includes, but is not limited to:
258
 
259
+ - Generating fake news, misinformation, or propaganda
260
+ - Promoting hate speech, discrimination, or violence against individuals or groups
261
+ - Impersonating individuals or organizations without their consent
262
+ - Engaging in cyberbullying or harassment
263
+ - Defamatory content
264
+ - Spamming or scamming
265
+ - Sharing confidential or sensitive information without proper authorization
266
+ - Violating the terms of use of the model or the data used to train it
267
+ - Creating automated bots for malicious purposes such as spreading malware, phishing scams, or spamming
268