jipso 0.1.18

JIPSO framework for prompt orchestration and comparison

📰 Paper

INSTALL

pip install jipso

🚀 QUICK START

Example 1

import jipso
import os

os.environ['OPENAI_API_KEY'] = 'sk-proj-...'

prompt1 = 'Write leave request email'
prompt2 = 'Write formal leave request email with clear reason and timeline'
o_eval = jipso.pvp(prompt1, prompt2)
print(o_eval)

# ✅ **Function executed:** pvp("Write leave request email", "Write formal leave request email with clear reason and timeline"

# **Test Input Generated:** Employee needs 3 days off next week for medical appointment

# **Results:**
# - **P1 Output:** Generic leave request mentioning time off needed
# - **P2 Output:** Structured email with specific dates, medical reason, coverage arrangements, and professional formatting

# 📊 **Score:** P1 = 3.2/10 (P2 baseline = 5.0)
# 📝 **Reasoning:** P1 produces vague, incomplete emails missing key details like specific dates, reasons, and professional structure. P2's explicit requirements for "clear reason and timeline" generate comprehensive, actionable requests that managers can easily approve. P2 consistently outperforms P1 in completeness, professionalism, and practical utility.

Example 2 (2235)

import jipso

os.environ['ANTHROPIC_API_KEY'] = 'sk-ant-...'

p1 = jipso.Prompt('Collect sales figures this week')
p1.add('Customer trend analysis')
print(p1)
# Collect sales figures this week and perform customer trend analysis

p2 = jipso.Prompt('Customer trend analysis')
p = p1 | p2
print(p)
# Collect sales figures this week and perform customer trend analysis

print(p > p2)
# True

p3 = p.enhance()
print(p3)
# Collect detailed sales figures for this week including revenue, units sold, and transaction counts by product category and customer segment, then perform comprehensive customer trend analysis identifying purchasing patterns, seasonal variations, and emerging opportunities with actionable insights and recommendations

print(set(p3))
# {
#   'Collect detailed sales figures for this week',
#   'Include revenue data', 
#   'Include units sold data',
#   'Include transaction counts',
#   'Categorize by product category',
#   'Categorize by customer segment', 
#   'Perform comprehensive customer trend analysis',
#   'Identify purchasing patterns',
#   'Identify seasonal variations', 
#   'Identify emerging opportunities',
#   'Provide actionable insights',
#   'Provide recommendations'
# }

print(dict(p3))
# {
#   "name": "comprehensive_sales_analysis",
#   "description": "Collect detailed sales data and perform customer trend analysis",
#   "data_collection": {
#     "timeframe": "this week",
#     "metrics": ["revenue", "units_sold", "transaction_counts"],
#     "segmentation": ["product_category", "customer_segment"]
#   },
#   "analysis": {
#     "type": "comprehensive_customer_trend_analysis",
#     "focus_areas": ["purchasing_patterns", "seasonal_variations", "emerging_opportunities"]
#   },
#   "output": {
#     "format": ["actionable_insights", "recommendations"],
#     "detail_level": "comprehensive"
#   }
# }

Example 3

import jipso

os.environ['GEMINI_API_KEY'] = 'sk-ant-...'

j = jipso.Judgement('models/gemini-1.5-flash')
i = 'Hi, I would like to ask about the Dell XPS 13 laptop. What is the current price and are there any promotions? Thank you!'
p = 'Please categorize this email into one of the following categories: Product Advice, Complaints, Technical Support, Orders, Other'
s = 'Based on the main content and purpose of the email. Choose only 1 category that best fits.'
o = j(i=i, p=p, s=s)
print(o)

# **Category: Product Advice**
# Reason: The email asks about current pricing and promotions for the Dell XPS 13 laptop, indicating the sender is researching to make a purchase decision — in the product consulting group.

Example 4

import jipso

os.environ['ALIBABACLOUD_API_KEY'] = 'sk-...'

compute = jipso.Compute(
  j = 'qwen-turbo',
  i = 'Hi, I would like to ask about the Dell XPS 13 laptop. What is the current price and are there any promotions? Thank you!',
  p = 'Please categorize this email into one of the following categories: Product Advice, Complaints, Technical Support, Orders, Other',
)
compute.s = 'Based on the main content and purpose of the email. Choose only 1 category that best fits.'
compute.exe()
print(compute.o)

🕌 ARCHITECT jipso-stack

Node	Docker Image	Engine	Role
Client Node	-	jipso-py	Request jipso.Compute.exe()
Worker Node	jipsofoundation/jipso	celery	Run jipso.Compute, wrap all AI model
Proxy Node	nginx	Nginx	Rate limiting
Cache Node	-	Redis GPU? (please build it or we will jipso-cache)	Cache VRAM
Broker Node	bitnami/kafka	Kafka	Message Queue Broker
Database Node	postgres	PostgreSQL	Database for jipso.Compute
Storage Node	minio/minio	Minio, S3, CDN	Media content
Metric Node	influxdb	InfluxDB	Metric: cost, SLA. Metric database and monitoring. Worker Node proactive push
Auth Node	keycloak/keycloak	Keycloak	Authentication, API key management

💰 SPONSORSHIP

This project has received no external funding, sponsorship, or investment. All development is fully volunteer-based at this stage.