same problem.
iPad Pro 11 with Apple Smart Keyboard Folio.
iPad OS 14.5.1

Pythonista can run on MacBook Air M1, and it's fast then iPad Pro 11 ( maybe 120% speed up).

If you are using Stash pip install tinyOpenAI. you can run it on Stash using: tinyopenai , run a cmd line ChatGPT.

tinyOpenAI for Pythonista Github

Features

• OpenAI ChatGPT and Whisper API library written in pure Python, so it can run in Python environment on M1/M2 Mac, iPad/iPhone(e.g. Pythonista, Juno, CODE, Pyto, ...), Android.
• Supports methods that conform to the ChatGPT API JSON format for API calls. Provides an easy-to-use quick dialog method, support for contextual associations; and easy language translation method.
• Support for Whisper interface calls to recognize and parse uploaded audio files as text messages or translate them into English.

Install

method 1: pip

• open StaSh, and then pip install tinyOpenAI

method 2: Copy code

• open tinyOpenAI Github ,and found tinyOpenAI.py, select all code, and copy to pythonista, run it.

example for ChatGPT

import tinyOpenAI

g = tinyOpenAI.ChatGPT('your OpenAI API_Key')
# g = tinyOpenAI.ChatGPT('your OpenAI API_Key','http://192.168.3.2:3128', Model='gpt-3.5-turbo-0301',Debug=True)
# Conversation
print( g.query('Write a rhyming poem with the sea as the title', system='You are a master of art, answer questions with emoji icons') )
# Continuous dialogue
print('======== continuous dialogue ============')
print(g.query('charles has $500, tom has $300, how much money do they have in total', True, 6))
print(g.query('charles and Tom who has more money', True, 6))
print(g.query('Sort them in order of money', True, 6))
# print history
print(g.Hinfo)
# clear Histroy
g.cHinfo()
# Statistics 
print('Call cnt: %d, Total using tokens: %d' % (g.Call_cnt, g.Total_tokens) )

example for Whisper

import tinyOpenAI

w = tinyOpenAI.Whisper('your OpenAI API_Key', Debug=True)
print(w.call('test1.m4a')) # or mp3/mp4 file
print(w.call('test2.m4a', 1)) # or mp3/mp4 file
print('Call cnt: %d, Total Texts: %d' % (w.Call_cnt, w.Total_tokens) )

It's running on my MacBook Air M1.
It's all like on my iPad Pro. but faster.

Just fork a bye-bye Feedly project ,and add some function. it’s can running on Pythonista. https://github.com/wolf71/bye-bye-feedly

1. download all your feedly account saved feeds to a json file.
2. run the websrv.py on pythonista .
3. it’s will open a web view to read all your feeds.

Traceback (most recent call last):
File "/var/containers/Bundle/Application/C653EC7F-B29A-4BAD-8AA1-1B562C926512/Pythonista3.app/Frameworks/PythonistaKit.framework/pylib/site-packages/scene.py", line 183, in _draw
if self.fixed_time_step and self.view:
AttributeError: 'MyScene' object has no attribute 'fixed_time_step'

@zrzka
Black Mamba short-cut can't work on Pythonista V3.4

@omz
keyboard shortcut error:

1. shift+tab not work.
2. cmd+/ not work.

@wolf71 update to V0.13

add openAI stream support. just set stream=True

Now you don't have to wait for the results to come out all at once after a long time, instead, you can quickly watch it pop up bit by bit.

import tinyOpenAI

g = tinyOpenAI.ChatGPT('your OpenAI API_Key', stream=True))

g.query('Write a rhyming poem with the sea as the title', system='You are a master of art, answer questions with emoji icons')

@rb HTTP Status_code 429
Rate limit errors ('Too Many Requests', ‘Rate limit reached’) are caused by hitting your organization's rate limit which is the maximum number of requests and tokens that can be submitted per minute. If the limit is reached, the organization cannot successfully submit requests until the rate limit is reset

update to V0.12

• Support Embedding API call, Embedding vectorization of the incoming text, support string or text array

Embedding (get the embedding vector of the text)

• init(self, API_Key='', Proxy='', Model='text-embedding-ada-002', URL='https://api.openai.com/v1/embeddings', Debug=False)
  • Initialize the creation of the Embedding object, with the following parameters
  • API_Key: your openAI API Key
  • Proxy: If needed, set your http proxy server, e.g.: http://192.168.3.1:3128
  • Model: If needed, you can set it according to the OpenAI API documentation.
  • URL: If OpenAI changes the API call address, you can change it here. Note that this is a list of two addresses, the first address is the original language output, the second address is translated into English.
  • Debug: if there is a network error or call error, whether to print out the error message, the default is not
• embed(data)
  • data: the string or list of strings to be encoded
  • The result is a list of embed vectors (1536 dimensions) corresponding to the strings, which can be obtained by
    • For the input string, ret[0].get('embedding') can be used to get the vector
    • For a list of input strings, you can get the list of vectors with [i.get('embedding') for i in ret]
• Statistics
  • Call_cnt: the cumulative number of calls to Whisper
  • Total_tokens: cumulative number of transcribed texts (Note: OpenAI is billed for the length of the audio, not the number of texts)
• Simple example

import tinyOpenAI

Embedding('your OpenAI API_Key', Debug=True)
r = e.embed('just for fun')
print('vector dimension:',len(r[0].get('embedding')))
# Compare the similarity of two texts
r = e.embed(['just for fun','hello world.'])
import numpy as np
print('Similarity result:',np.dot(r[0].get('embedding'), r[1].get('embedding')))

Ref:

• OpenAI Embeddings Doc

If you are using Stash pip install tinyOpenAI. you can run it on Stash using: tinyopenai , run a cmd line ChatGPT.

tinyOpenAI for Pythonista Github

Features

• OpenAI ChatGPT and Whisper API library written in pure Python, so it can run in Python environment on M1/M2 Mac, iPad/iPhone(e.g. Pythonista, Juno, CODE, Pyto, ...), Android.
• Supports methods that conform to the ChatGPT API JSON format for API calls. Provides an easy-to-use quick dialog method, support for contextual associations; and easy language translation method.
• Support for Whisper interface calls to recognize and parse uploaded audio files as text messages or translate them into English.

Install

method 1: pip

• open StaSh, and then pip install tinyOpenAI

method 2: Copy code

• open tinyOpenAI Github ,and found tinyOpenAI.py, select all code, and copy to pythonista, run it.

example for ChatGPT

import tinyOpenAI

g = tinyOpenAI.ChatGPT('your OpenAI API_Key')
# g = tinyOpenAI.ChatGPT('your OpenAI API_Key','http://192.168.3.2:3128', Model='gpt-3.5-turbo-0301',Debug=True)
# Conversation
print( g.query('Write a rhyming poem with the sea as the title', system='You are a master of art, answer questions with emoji icons') )
# Continuous dialogue
print('======== continuous dialogue ============')
print(g.query('charles has $500, tom has $300, how much money do they have in total', True, 6))
print(g.query('charles and Tom who has more money', True, 6))
print(g.query('Sort them in order of money', True, 6))
# print history
print(g.Hinfo)
# clear Histroy
g.cHinfo()
# Statistics 
print('Call cnt: %d, Total using tokens: %d' % (g.Call_cnt, g.Total_tokens) )

example for Whisper

import tinyOpenAI

w = tinyOpenAI.Whisper('your OpenAI API_Key', Debug=True)
print(w.call('test1.m4a')) # or mp3/mp4 file
print(w.call('test2.m4a', 1)) # or mp3/mp4 file
print('Call cnt: %d, Total Texts: %d' % (w.Call_cnt, w.Total_tokens) )

For Bioinformatics Support

• loadcsv also support tsv format file (\t split type), just using:
  • loadcsv test.tsv tb1 1
• loadcsv also support bioinformatics .maf/.vcf/.sam/.gtf/.gff/.gpd file, just using:
  • loadcsv test.maf tb1 or loadcsv test.vcf tb1
• Load GenBank Format Features Data.
  • using: loadgb gbfile tb01
  • load bioinformatics GenBank file (etc: .gb/.gbff/.gpff) features info to table.

Crawl website data, parse and write to database

• using: loop loadweb url=https://xxx.xxx.com/xxx re=(\d+), base on python regular module, Qsqlite loop nesting function, can achieve complex web data crawling and content extraction, and then write to the database.

All new function please check the Readme file

Qsqlite is one py file only, you can just copy the Qsqlite.py file on Github to Pythonista, and then run it on python3 mode.

What

• AntPool provides a server and resource client that makes it easy and fast to set up a multi-computer distributed environment.
• AntPool provides a development library consistent with Python's native concurrent.futures.Executor asynchronous library, allowing you to quickly deploy your code to run in the multi-computer distributed environment.

Why

• In some application scenarios, we want programs to run on multiple machines in parallel to gain performance, network access improvements;
  • Computational Performance Boost: For complex computation and analysis in Python, multi-threaded Python can't boost on multi-core CPUs(limit by GIL), so Python introduced the concurrent.futures.ProcessPoolExecutor, Multi-process development is easy to implement; but when higher performance is needed, a multi-computer distributed environment quickly allows you to change just a few lines of code and take performance to new heights.
  • Network Capability Improvement: Crawling data on the Internet, analyzing and extracting it, is often limited by the traffic of the crawled server. Therefore, it is necessary to deploy the program to multiple machines with different network addresses for crawling, which is tedious and error-prone to do manually, but a multi-machine distributed environment can quickly solve such problems.
• Everyone Can Using Instead of requiring complex configurations, specialized equipment, and relearning how to develop, AntPool is available to everyone.
  • Any single device can be used as a server or resource client or both. These include PC/Mac/Linux computers, and even devices like your iPhone/iPad/Android that can run Python.
  • You can use all your devices at home to help you improve your computing performance, or you can ask your colleagues at work to contribute some of their CPU cores to help you achieve complex computations.
  • All computing resources are connected to a server-driven cluster network by running a resource client, which can be joined and quit flexibly.
  • No need to learn new concepts, as you only need to understand Python's native concurrent.futures.Executor, as the AntPool classes are same.
• Mobile Device Support iPad is a great tool, and there are already many great apps that run Python and iPython Notebook, but due to iOS security restrictions, many libraries can't be installed, Python also don't support multi-processing. With AntPool, you can call a multi-machine cluster directly on the iPad and run features on the cluster that are not available on the iPad or get more performance.
  • iPad Python Apps
    • Pythonista(requires stash , for pip support)
    • Pyto
    • CodeApp
  • iPad iPython Notebook/Jupyter Apps
    • Juno
    • Carnets

How

• Runtime requirements:
  • Python 3.5 or above
• Quick Start:
  • pip3 install antpool
  • for Pythonista user, please pip install tornado sorry for a bug, need Manual install tornado.
  • Standalone deployment server and resource client
    • Run in a terminal environment: antpoolsrv /v
    • Open a new terminal and run: antpoolclient ws://127.0.0.1:8086
    • If you need to access more resource clients to improve cluster performance, you can do so on other machines:
      • pip install antpool
      • antpoolclient ws://server IP:8086
    • Each antpoolclient client running on the same machine will use one CPU core, so if you want to exploit the performance of the whole machine, you can run multiple clients (e.g. for a 4-core CPU machine, you can run 4 clients)
  • Run the demo program
    • python3 demo1.py (demo1.py on github.com/wolf71/AntPool/demo)
  • Structure diagram
    

!!! Security Tip !!!

• The resource client is execution of the submitted modules, so the device running the resource client will be vulnerable to attacks by users with bad intentions, the system does not perform security checks on the executed modules, so it is recommended that the user running the resource client needs to verify that the user is trustworthy;
• For the sake of simplicity, the entire system currently does not use user/password authentication mechanisms (interfaces are reserved in the code); therefore, it is recommended that the server be deployed on the public network with security in mind;

User Manual

check it on GitHub

Add iPython Notebook demo.

You can using iPython Notebook on iPad with JUNO or Carnets.

check it on GitHub