प्रॉम्प्ट डिज़ाइन से जुड़ी रणनीतियां

इस पेज पर, डिज़ाइन से जुड़ी कुछ सामान्य रणनीतियों के बारे में बताया गया है, जिनका इस्तेमाल प्रॉम्प्ट डिज़ाइन करते समय किया जा सकता है. वैसे तो प्रॉम्प्ट डिज़ाइन करने का कोई सही या गलत तरीका नहीं है, लेकिन कुछ सामान्य रणनीतियां हैं, जिनका इस्तेमाल करके मॉडल के जवाबों पर असर डाला जा सकता है. कड़े टेस्ट और इवैलुएशन से, मॉडल की परफ़ॉर्मेंस को ऑप्टिमाइज़ करने में मदद मिलती है.

बड़े लैंग्वेज मॉडल (एलएलएम) को बहुत सारे टेक्स्ट डेटा पर ट्रेनिंग दी जाती है, ताकि भाषा की इकाइयों के बीच के पैटर्न और इनके बीच के संबंध को समझा जा सके. कुछ टेक्स्ट (प्रॉम्प्ट) दिए जाने पर, भाषा मॉडल यह अनुमान लगा सकते हैं कि आने वाले समय में क्या हो सकता है. उदाहरण के लिए, अपने-आप पूरा होने वाला एक बेहतर टूल. इसलिए, प्रॉम्प्ट डिज़ाइन करते समय, उन अलग-अलग बातों का ध्यान रखें जो इस बात पर असर डाल सकती हैं कि मॉडल के अगले चरण क्या हैं.

साफ़ तौर पर और सटीक निर्देश दें

मॉडल के व्यवहार को पसंद के मुताबिक बनाने का एक असरदार और असरदार तरीका है, ताकि हम मॉडल को इसके बारे में निर्देश दे सकें कि क्या करना है. पक्का करें कि आपके दिए गए निर्देश साफ़ और सटीक हों. निर्देश, चरण-दर-चरण निर्देशों की सूची जितने आसान या उपयोगकर्ता के अनुभव और सोच को बताने जितना मुश्किल हो सकते हैं.

तय करें कि आपको क्या करना है

उस टास्क के बारे में पूरी जानकारी दें जिसे आपको मॉडल से पूरा करना है. यह काम, निर्देशों की सिलसिलेवार निर्देशों की सूची जितना आसान हो सकता है. इसके अलावा, यह उपयोगकर्ता के अनुभव और मानसिकता को मैप करने जितना मुश्किल भी हो सकता है. इस प्रॉम्प्ट से मॉडल को टेक्स्ट के एक हिस्से की खास जानकारी देने के लिए कहा जाता है:

Summarize this text.
Text: A quantum computer exploits quantum mechanical phenomena to perform calculations exponentially
faster than any modern traditional computer. At very tiny scales, physical matter acts as both
particles and as waves, and quantum computing uses specialized hardware to leverage this behavior.
The operating principles of quantum devices is beyond the scope of classical physics. When deployed
at scale, quantum computers could be used in a wide variety of applications such as: in
cybersecurity to break existing encryption methods while helping researchers create new ones, in
meteorology to develop better weather forecasting etc. However, the current state of the art quantum
computers are still largely experimental and impractical.
  
Quantum computers utilize quantum mechanical phenomena to perform computations significantly faster
than traditional computers. By exploiting the particle-wave duality of matter at tiny scales,
quantum computers leverage specialized hardware to solve problems beyond classical physics.
Potential applications include breaking encryption, developing better weather forecasting, and more.
However, current quantum computers are still largely experimental and impractical.
  

कोई सीमा तय करें

प्रॉम्प्ट पढ़ने या जवाब जनरेट करने से जुड़ी कोई समस्या बताएं. मॉडल को यह बताया जा सकता है कि उसे क्या करना है और क्या नहीं. पिछले उदाहरण में दिया गया उदाहरण, ओरिजनल टेक्स्ट से छोटा है, लेकिन हो सकता है कि आपको जवाब को छोटा रखना हो. इस मामले में, प्रॉम्प्ट में इस बारे में एक कंस्ट्रेंट तय किया जा सकता है कि जवाब को कितने समय तक रखना है:

Summarize this text in two sentences.
Text: A quantum computer exploits quantum mechanical phenomena to perform calculations exponentially
faster than any modern traditional computer. At very tiny scales, physical matter acts as both
particles and as waves, and quantum computing uses specialized hardware to leverage this behavior.
The operating principles of quantum devices is beyond the scope of classical physics. When deployed
at scale, quantum computers could be used in a wide variety of applications such as: in
cybersecurity to break existing encryption methods while helping researchers create new ones, in
meteorology to develop better weather forecasting etc. However, the current state of the art quantum
computers are still largely experimental and impractical.
    

Quantum computers use quantum mechanical phenomena to perform calculations much faster than
traditional computers. They have the potential to revolutionize various fields, but current quantum
computers are still experimental and impractical.
    

रिस्पॉन्स का फ़ॉर्मैट तय करना

जवाब का फ़ॉर्मैट बताने वाले निर्देश भी दिए जा सकते हैं. उदाहरण के लिए, जवाब को टेबल, बुलेट वाली सूची, एलिवेटर पिच, कीवर्ड, वाक्य या पैराग्राफ़ के तौर पर फ़ॉर्मैट करने के लिए कहा जा सकता है. यह प्रॉम्प्ट, मॉडल को टेक्स्ट में मुख्य जानकारी के बुलेट पॉइंट देने का निर्देश देता है:

Summarize this text as bullets points of key information.
Text: A quantum computer exploits quantum mechanical phenomena to perform calculations exponentially
faster than any modern traditional computer. At very tiny scales, physical matter acts as both
particles and as waves, and quantum computing uses specialized hardware to leverage this behavior.
The operating principles of quantum devices is beyond the scope of classical physics. When deployed
at scale, quantum computers could be used in a wide variety of applications such as: in
cybersecurity to break existing encryption methods while helping researchers create new ones, in
meteorology to develop better weather forecasting etc. However, the current state of the art quantum
computers are still largely experimental and impractical.
  
- Quantum computers leverage quantum mechanical phenomena for exponentially faster calculations than
  traditional computers.
- Quantum devices utilize specialized hardware to exploit tiny-scale matter behavior as both
  particles and waves.
- Their operating principles transcend classical physics.
- Potential applications include cybersecurity, weather forecasting, and more.
- Current quantum computers are largely experimental and impractical.
  

खास जानकारी

  • मॉडल को निर्देश दें कि क्या करना है.
  • निर्देशों को साफ़ तौर पर और सटीक बनाएं.
  • आउटपुट के लिए कोई सीमा या फ़ॉर्मैटिंग की ज़रूरी शर्तें बताएं.

शॉट के उदाहरण शामिल करें

आपके पास प्रॉम्प्ट में ऐसे उदाहरण शामिल करने का विकल्प होता है जो मॉडल को दिखाते हैं कि उसे कैसे ठीक किया जा रहा है. मॉडल, उदाहरणों से पैटर्न और संबंधों की पहचान करता है और जवाब जनरेट करते समय उन्हें लागू करता है. जिन प्रॉम्प्ट में कुछ उदाहरण होते हैं उन्हें फ़्यू-शॉट प्रॉम्प्ट कहा जाता है, जबकि ऐसे प्रॉम्प्ट जिनमें कोई उदाहरण नहीं है उन्हें ज़ीरो-शॉट प्रॉम्प्ट कहा जाता है. कुछ-कुछ प्रॉम्प्ट का इस्तेमाल अक्सर मॉडल रिस्पॉन्स की फ़ॉर्मैटिंग, फ़्रेज़िंग, स्कोपिंग या सामान्य पैटर्निंग को रेगुलेट करने के लिए किया जाता है. खास और अलग-अलग उदाहरणों का इस्तेमाल करें, ताकि मॉडल को अपना फ़ोकस सटीक बनाने और ज़्यादा सटीक नतीजे जनरेट करने में मदद मिल सके.

हमारा सुझाव है कि आप हमेशा अपने प्रॉम्प्ट में, कुछ शॉट के उदाहरण शामिल करें. जिन प्रॉम्प्ट में कुछ उदाहरण न हों वे कम असरदार होते हैं, क्योंकि वे इस मॉडल को बताते हैं कि निर्देशों को कैसे लागू किया जाए. असल में, अगर आपके उदाहरण में टास्क साफ़ तौर पर दिख रहा है, तो अपने प्रॉम्प्ट से निर्देश हटाए जा सकते हैं.

ज़ीरो-शॉट बनाम कुछ-शॉट प्रॉम्प्ट

नीचे शून्य-शॉट प्रॉम्प्ट, मॉडल से सबसे सही जानकारी चुनने के लिए कहता है.

Please choose the best explanation to the question:

Question: How is snow formed?
Explanation1: Snow is formed when water vapor in the air freezes into ice crystals in the
atmosphere, which can combine and grow into snowflakes as they fall through the atmosphere and
accumulate on the ground.
Explanation2: Water vapor freezes into ice crystals forming snow.
Answer:
  
Explanation1
  

अगर आपके इस्तेमाल के उदाहरण के लिए मॉडल को छोटे जवाब देने की ज़रूरत है, तो आप प्रॉम्प्ट में ऐसे उदाहरण शामिल कर सकते हैं जो छोटे जवाबों को प्राथमिकता देते हैं.

इस प्रॉम्प्ट में दो उदाहरण दिए गए हैं, जो छोटे एक्सप्लेनेशंस को प्राथमिकता देते हैं. इस जवाब में, आपको दिख सकता है कि उदाहरणों ने मॉडल को पहले की तरह लंबे एक्सप्लेनेशन (एक्सप्लेनेशंस1) के बजाय, छोटा ब्यौरा (एक्सप्लेनेशंस2) चुनने का निर्देश दिया है.

Please choose the best explanation to the question:

Question: Why is sky blue?
Explanation1: The sky appears blue because of Rayleigh scattering, which causes shorter blue
wavelengths of light to be scattered more easily than longer red wavelengths, making the sky look
blue.
Explanation2: Due to Rayleigh scattering effect.
Answer: Explanation2

Question: What is the cause of earthquakes?
Explanation1: Sudden release of energy in the Earth's crust.
Explanation2: Earthquakes happen when tectonic plates suddenly slip or break apart, causing a
release of energy that creates seismic waves that can shake the ground and cause damage.
Answer: Explanation1

Question: How is snow formed?
Explanation1: Snow is formed when water vapor in the air freezes into ice crystals in the
atmosphere, which can combine and grow into snowflakes as they fall through the atmosphere and
accumulate on the ground.
Explanation2: Water vapor freezes into ice crystals forming snow.
Answer:
  
Explanation2
  

उदाहरणों की अधिकतम संख्या का पता लगाना

सबसे ज़्यादा मनमुताबिक नतीजे पाने के लिए, प्रॉम्प्ट में उदाहरणों की संख्या के साथ प्रयोग किया जा सकता है. PaLM और Gemini जैसे मॉडल अक्सर कुछ उदाहरणों का इस्तेमाल करके, पैटर्न हासिल कर सकते हैं. हालांकि, आपको यह देखना पड़ सकता है कि कितने उदाहरणों से मनमुताबिक नतीजे मिलेंगे. BERT जैसे आसान मॉडल के लिए, आपको और उदाहरणों की ज़रूरत पड़ सकती है. साथ ही, अगर आपने बहुत सारे उदाहरण शामिल किए हैं, तो मॉडल, दिए गए उदाहरणों के जवाब को ओवरफ़िट करना शुरू कर सकता है.

एंटीपैटर्न के बजाय पैटर्न दिखाने के लिए, उदाहरण इस्तेमाल करें

अगर आपको उदाहरणों से बचना है, तो उदाहरण के बजाय एंटीपैटर्न दिखाने वाले मॉडल को दिखाने के लिए, उदाहरणों का इस्तेमाल करना होगा.

नेगेटिव पैटर्न:

Don't end haikus with a question:
Haiku are fun
A short and simple poem
Don't you enjoy them?
  

पॉज़िटिव पैटर्न:

Always end haikus with an assertion:
Haiku are fun
A short and simple poem
A joy to write
  

सभी उदाहरणों में एक जैसी फ़ॉर्मैटिंग का इस्तेमाल करना

अनचाहे फ़ॉर्मैट वाले जवाबों से बचने के लिए पक्का करें कि कुछ शॉट के उदाहरणों का स्ट्रक्चर और फ़ॉर्मैटिंग एक जैसी हो. प्रॉम्प्ट में कुछ शॉट के उदाहरण जोड़ने का एक मुख्य मकसद, मॉडल को रिस्पॉन्स फ़ॉर्मैट दिखाना है. इसलिए, यह ज़रूरी है कि सभी उदाहरणों के लिए, एक जैसा फ़ॉर्मैट इस्तेमाल किया जा सके. खास तौर पर, एक्सएमएल टैग, व्हाइट स्पेस, नई लाइन, और उदाहरण स्प्लिटर पर ध्यान देना ज़रूरी है.

खास जानकारी

  • प्रॉम्प्ट में तुरंत जवाब देने के उदाहरण शामिल करने से, मॉडल को जवाब देने का तरीका जानने में मदद मिलती है.
  • पैटर्न के उदाहरणों से बचने के बजाय, उन पैटर्न के उदाहरण दें जिन्हें फ़ॉलो किया जाना चाहिए.
  • शामिल किए जाने वाले प्रॉम्प्ट की संख्या के साथ एक्सपेरिमेंट करें. मॉडल के आधार पर, कुछ उदाहरण मॉडल के काम करने के तरीके को बदलने में कोई असर नहीं डालते. बहुत ज़्यादा उदाहरणों की वजह से मॉडल ज़रूरत से ज़्यादा फ़िट हो जाता है.
  • सभी उदाहरणों में एक जैसी फ़ॉर्मैटिंग का इस्तेमाल करना

काम की जानकारी जोड़ना

मॉडल में सभी ज़रूरी जानकारी मौजूद होने के बजाय, मॉडल को समस्या हल करने के लिए निर्देश और जानकारी दी जा सकती है. काम की इस जानकारी की मदद से, मॉडल यह समझ पाता है कि उसे क्या करना है और उसे क्या करना है.

इस उदाहरण में, मॉडल से राऊटर के लिए समस्या हल करने से जुड़े दिशा-निर्देश देने के लिए कहा गया है:

What should I do to fix my disconnected wifi? The light on my Google Wifi router is yellow and
blinking slowly.
  
1. **Check the Cables:**
- Make sure that all cables are securely connected to your Google Wifi router and modem.
- Try replacing the cables with new ones to ensure they are not faulty.


2. **Reboot the Router:**
- Unplug your Google Wifi router and modem from the power outlet.
- Wait for at least 30 seconds.
- Plug the modem and router back in and wait for them to start up.


3. **Factory Reset the Router:**
- Resetting the router will restore it to its default settings.
- Find the small reset button on the back of the router.
- Press and hold the reset button with a paper clip or similar object for 10 seconds.
- Release the button and wait for the router to restart.


4. **Check Your Internet Connection:**
- Use a wired connection to connect a device directly to your modem.
- If the internet works, the issue may be with your router.
- If the internet still doesn't work, contact your Internet Service Provider (ISP).


5. **Update Router's Firmware:**
- Check if there is a firmware update available for your Google Wifi router.
- Download and install the latest firmware update from the Google Wifi app.


6. **Interference and Obstacles:**
- Make sure the router is not near electronic devices that can cause interference, like microwaves or cordless phones.
- Remove any obstacles between the router and your devices, such as furniture or thick walls.


7. **Change Wi-Fi Channel:**
- In the Google Wifi app, go to Settings > Network & General > Advanced Networking.
- Try changing the Wi-Fi channel to a less congested one.


8. **Contact Google Support:**
- If you've tried the above steps and the issue persists, contact Google support for further assistance.
  

जवाब, सामान्य समस्या हल करने वाली जानकारी की तरह दिखता है जो खास तौर पर राऊटर या एलईडी इंडिकेटर लाइटों की स्थिति के बारे में नहीं है.

किसी खास राऊटर के लिए, जवाब को अपनी पसंद के मुताबिक बनाने के लिए, जवाब देते समय राऊटर की समस्या हल करने वाली गाइड देखें. इस गाइड का इस्तेमाल जवाब देते समय किया जा सकता है.

Answer the question using the text below. Respond with only the text provided.
Question: What should I do to fix my disconnected wifi? The light on my Google Wifi router is yellow and blinking slowly.

Text:
Color: Slowly pulsing yellow
What it means: There is a network error.
What to do:
Check that the Ethernet cable is connected to both your router and your modem and both devices are turned on. You might need to unplug and plug in each device again.

Color: Fast blinking yellow
What it means: You are holding down the reset button and are factory resetting this device.
What to do:
If you keep holding down the reset button, after about 12 seconds, the light will turn solid yellow. Once it is solid yellow, let go of the factory reset button.

Color: Solid yellow
What it means: Router is factory resetting.
What to do:
This can take up to 10 minutes. When it's done, the device will reset itself and start pulsing white, letting you know it's ready for setup.

Color: Solid red
What it means: Something is wrong.
What to do:
Critical failure. Factory reset the router. If the light stays red, contact Wifi customer support.
  
Check that the Ethernet cable is connected to both your router and your modem and both devices are turned on. You might need to unplug and plug in each device again.
  

खास जानकारी

  • प्रॉम्प्ट में ऐसी जानकारी (कॉन्टेक्स्ट) भी शामिल करें जिसका इस्तेमाल आपको जवाब जनरेट करने के लिए करना है.
  • मॉडल को निर्देश दें कि कॉन्टेक्स्ट के हिसाब से जानकारी का इस्तेमाल कैसे किया जाए.

प्रीफ़िक्स जोड़ें

प्रीफ़िक्स एक ऐसा शब्द या वाक्यांश है जिसे प्रॉम्प्ट के कॉन्टेंट में जोड़ा जाता है. इससे कई काम किए जा सकते हैं. यह इस बात पर निर्भर करता है कि प्रीफ़िक्स कहां लगाया गया है:

  • इनपुट प्रीफ़िक्स: मॉडल के इनपुट सिग्नल के शब्दों के हिसाब से सही हिस्सों में प्रीफ़िक्स जोड़ना. उदाहरण के लिए, "अंग्रेज़ी:" और "फ़्रेंच:" प्रीफ़िक्स, दो अलग-अलग भाषाओं को तय करते हैं.
  • आउटपुट प्रीफ़िक्स: भले ही आउटपुट, मॉडल से जनरेट होता है, लेकिन प्रॉम्प्ट में आउटपुट के लिए प्रीफ़िक्स जोड़ा जा सकता है. आउटपुट प्रीफ़िक्स, मॉडल को इस बारे में जानकारी देता है कि रिस्पॉन्स के तौर पर क्या मिल सकता है. उदाहरण के लिए, आउटपुट प्रीफ़िक्स "JSON:" मॉडल से पता चलता है कि आउटपुट JSON फ़ॉर्मैट में होना चाहिए.
  • उदाहरण प्रीफ़िक्स: कुछ शॉट वाले प्रॉम्प्ट में, उदाहरणों में प्रीफ़िक्स जोड़ने से ऐसे लेबल मिलते हैं जिनका इस्तेमाल मॉडल, आउटपुट जनरेट करते समय कर सकता है. इससे आउटपुट कॉन्टेंट को पार्स करना आसान हो जाता है.

यहां दिए गए उदाहरण में, इनपुट प्रीफ़िक्स "Text:" है और "जवाब है:" आउटपुट प्रीफ़िक्स है.

Classify the text as one of the following categories.
- large
- small
Text: Rhino
The answer is: large
Text: Mouse
The answer is: small
Text: Snail
The answer is: small
Text: Elephant
The answer is:
  
large
  

मॉडल को आंशिक इनपुट पूरा करने दें

जनरेटिव लैंग्वेज मॉडल, अपने-आप पूरा होने वाले एक ऐडवांस टूल की तरह काम करते हैं. जब आंशिक कॉन्टेंट उपलब्ध कराया जाता है, तो मॉडल बाकी कॉन्टेंट उपलब्ध करा सकता है. इसके अलावा, मॉडल के हिसाब से वह कॉन्टेंट, रिस्पॉन्स के तौर पर उपलब्ध हो सकता है. ऐसा करते समय, अगर कोई उदाहरण या कॉन्टेक्स्ट शामिल किया जाता है, तो मॉडल उन उदाहरणों या कॉन्टेक्स्ट को ध्यान में रख सकता है.

इस उदाहरण में, निर्देश और इकाई इनपुट के साथ प्रॉम्प्ट दिया गया है:

For the given order, return a JSON object that has the fields cheeseburger, hamburger, fries, or
drink, with the value being the quantity.

Order: A burger and a drink.
  
```
{
"cheeseburger": 0,
"hamburger": 1,
"fries": 0,
"drink": 1
}
```
  

हालांकि, मॉडल ने ऐसा ही करने के लिए कहा था, फिर भी निर्देशों को आम भाषा में लिखना कभी-कभी चुनौती भरा हो सकता है और मॉडल को समझने में बहुत मुश्किल होती है. उदाहरण के लिए, किसी रेस्टोरेंट के मेन्यू में कई आइटम हो सकते हैं. JSON के रिस्पॉन्स का साइज़ कम करने के लिए, शायद आप उन आइटम को हटाना चाहें जिन्हें क्रम से नहीं लगाया गया है. इस मामले में, एक उदाहरण और रिस्पॉन्स प्रीफ़िक्स दिया जा सकता है. इसके बाद, मॉडल को इसे पूरा करने दें:

Valid fields are cheeseburger, hamburger, fries, and drink.
Order: Give me a cheeseburger and fries
Output:
```
{
  "cheeseburger": 1,
  "fries": 1
}
```
Order: I want two burgers, a drink, and fries.
Output:
  
```
{
"hamburger": 2,
"drink": 1,
"fries": 1
}
```
  

ध्यान दें कि कैसे "चीज़बर्गर" को आउटपुट से बाहर रखा गया, क्योंकि वह ऑर्डर का हिस्सा नहीं था.

मॉडल को उसका रिस्पॉन्स फ़ॉर्मैट करने का निर्देश दें

जवाब देने की रणनीति से, जवाब को फ़ॉर्मैट करने में भी मदद मिल सकती है. इस उदाहरण में, मॉडल को निबंध की आउटलाइन बनाने के लिए कहा जाता है:

Create an outline for an essay about hummingbirds.
  
I. Introduction
A. Hook: Begin with an intriguing fact or description of hummingbirds.
B. Provide some background information on hummingbirds.
1. Mention their small size and long, thin beaks.
2. Highlight their unique ability to hover and fly backward.

II. Body
A. Anatomy and Physiology
1. Elaborate on their tiny size, making comparisons to other animals.
2. Describe their wings and how they enable hovering and swift flight.
3. Discuss their metabolism and energy requirements, emphasizing their caloric needs.

B. Feeding Habits and Diet
1. Explain their primary food source, nectar, and how they extract it from flowers.
2. Describe their ability to feed from a wide variety of flowers.
3. Mention their consumption of insects and the nutritional value they provide.

C. Migration Patterns
1. Discuss the incredible long-distance migrations of some hummingbird species.
2. Explain the phenomenon of seasonal migration and the environmental factors driving it.
3. Describe the routes taken by different species during migration.

D. Conservation and Threats
1. Highlight the importance of hummingbirds in ecosystems as pollinators.
2. Discuss the challenges and threats they face, such as habitat loss, nectar scarcity, and climate change.
3. Mention conservation efforts and ways to support hummingbird populations.

III. Conclusion
A. Summarize the main points covered in the essay.
B. Reinforce the importance of hummingbirds and their ecological role.
C. Provide a final thought or call to action to protect these fascinating creatures.
  

प्रॉम्प्ट में आउटलाइन का फ़ॉर्मैट नहीं बताया गया था, लेकिन मॉडल ने आपके लिए एक फ़ॉर्मैट चुना. मॉडल से किसी फ़ॉर्मैट में आउटलाइन तैयार कराने के लिए, आपके पास ऐसा टेक्स्ट जोड़ने का विकल्प होता है जो आउटलाइन की शुरुआत के बारे में बताता हो. साथ ही, मॉडल को आपके शुरू किए गए पैटर्न के हिसाब से पूरा करने दें.

Create an outline for an essay about hummingbirds.
I. Introduction
   *
  
I. Introduction
* Capture the reader's attention with an interesting anecdote or fact about hummingbirds.
* Provide a brief background on hummingbirds, including their unique characteristics.
* State the main points that will be covered in the essay.

II. The Unique Characteristics of Hummingbirds
* Discuss the hummingbird's exceptionally rapid wing-beat, allowing them to hover and fly backward.
* Highlight their tiny size, making them the smallest birds in the world.
* Describe their long, slender beaks and tongues, allowing them to reach deep into flowers to extract nectar.
* Explain their iridescent plumage, which creates brilliant and vibrant colors.

III. Hummingbird Migration and Distribution
* Discuss the extensive migration patterns of hummingbirds, covering vast distances across continents.
* Describe the different habitats where hummingbirds can be found, including tropical rainforests, deserts, and urban areas.
* Explain the seasonal changes that influence hummingbird migration and habitat selection.

IV. Hummingbird Diet and Feeding Habits
* Focus on the hummingbird's primary food source, which is nectar from flowers.
* Describe their ability to consume insects, spiders, and other small invertebrates for protein.
* Explain how hummingbirds use their long, slender tongues to extract nectar from flowers.
* Discuss the importance of flowering plants in providing sustenance for hummingbirds.

V. Reproduction and Nesting Behavior
* Describe the mating rituals and courtship behavior of hummingbirds.
* Explain the process of nest-building, including the materials used and the unique characteristics of hummingbird nests.
* Discuss the role of the female hummingbird in laying and incubating eggs.
* Describe the development of hummingbird chicks and the role of both parents in providing care.

VI. Threats to Hummingbirds and Conservation Efforts
* Highlight habitat loss, climate change, and deforestation as significant threats to hummingbird populations.
* Describe the role of habitat conservation initiatives in protecting hummingbird habitats.
* Explain the importance of creating hummingbird-friendly gardens and landscapes.
* Discuss the efforts of organizations and individuals working to conserve hummingbird populations.

VII. Conclusion
* Summarize the main points covered in the essay, emphasizing the unique characteristics, migration patterns, diet, reproduction, and conservation of hummingbirds.
* Provide a closing statement that leaves a lasting impression on the reader and reinforces the significance of these fascinating creatures.
  

खास जानकारी

  • अगर मॉडल को आंशिक इनपुट दिया जाता है, तो मॉडल उस इनपुट को प्रॉम्प्ट में उपलब्ध उदाहरण या संदर्भ के आधार पर पूरा करता है.
  • काम को आम भाषा में बताने के बजाय, मॉडल को इनपुट पूरा करना कभी-कभी ज़्यादा आसान होता है.
  • सवाल का अधूरा जवाब जोड़ने से, मॉडल को मनमुताबिक पैटर्न या फ़ॉर्मैट फ़ॉलो करने में मदद मिल सकती है.

प्रॉम्प्ट को आसान कॉम्पोनेंट में बांटें

जटिल प्रॉम्प्ट की ज़रूरत वाले इस्तेमाल के उदाहरणों के लिए, चीज़ों को आसान कॉम्पोनेंट में बांटकर, मॉडल को इस जटिलता को मैनेज करने में मदद की जा सकती है.

निर्देशों का विश्लेषण करें

एक ही प्रॉम्प्ट में कई निर्देश देने के बजाय, हर निर्देश के लिए एक प्रॉम्प्ट बनाएं. आपके पास यह चुनने का विकल्प है कि उपयोगकर्ता के इनपुट के आधार पर, किस प्रॉम्प्ट को प्रोसेस करना है.

चेन प्रॉम्प्ट

कई क्रम में चलने वाले मुश्किल टास्क के लिए, हर चरण को प्रॉम्प्ट बनाएं और प्रॉम्प्ट को क्रम से एक साथ जोड़ें. प्रॉम्प्ट के इस क्रम में, एक प्रॉम्प्ट का आउटपुट अगले प्रॉम्प्ट का इनपुट बन जाता है. क्रम में दिए गए आखिरी प्रॉम्प्ट का आउटपुट, फ़ाइनल आउटपुट होता है.

कुल जवाब

एग्रीगेशन का मतलब है कि जब आपको डेटा के अलग-अलग हिस्सों पर अलग-अलग काम करने हों और फ़ाइनल आउटपुट देने के लिए नतीजों को एग्रीगेट करना हो. उदाहरण के लिए, आपके पास मॉडल को डेटा के पहले हिस्से पर एक कार्रवाई करने, बाकी डेटा पर दूसरी कार्रवाई करने और नतीजे इकट्ठा करने का निर्देश देने का विकल्प होता है.

खास जानकारी

  • मुश्किल निर्देशों को एक प्रॉम्प्ट में बांटकर, हर निर्देश के लिए एक प्रॉम्प्ट बनाएं. साथ ही, यह तय करें कि उपयोगकर्ता के इनपुट के हिसाब से कौनसा प्रॉम्प्ट लागू करना चाहिए.
  • एक से ज़्यादा चरणों को अलग-अलग प्रॉम्प्ट में बांटें और उन्हें इस तरह से चेन करें कि पिछले प्रॉम्प्ट का आउटपुट, नीचे दिए गए प्रॉम्प्ट का इनपुट बन जाए.
  • साथ-साथ चलने वाले टास्क का बंटवारा करें और फ़ाइनल आउटपुट पाने के लिए जवाबों को एग्रीगेट करें.

अलग-अलग पैरामीटर वैल्यू के साथ प्रयोग करना

मॉडल को भेजे जाने वाले हर कॉल में पैरामीटर वैल्यू शामिल होती हैं, जो यह कंट्रोल करती हैं कि मॉडल कैसे रिस्पॉन्स जनरेट करता है. मॉडल, अलग-अलग पैरामीटर वैल्यू के लिए अलग-अलग नतीजे जनरेट कर सकता है. किसी टास्क के लिए सबसे सही वैल्यू पाने के लिए, अलग-अलग पैरामीटर वैल्यू के साथ प्रयोग करें. अलग-अलग मॉडल के लिए उपलब्ध पैरामीटर अलग-अलग हो सकते हैं. सबसे सामान्य पैरामीटर नीचे दिए गए हैं:

  • ज़्यादा से ज़्यादा आउटपुट टोकन
  • तापमान
  • टॉप-के
  • टॉप-पी

ज़्यादा से ज़्यादा आउटपुट टोकन

रिस्पॉन्स में जनरेट हो सकने वाले ज़्यादा से ज़्यादा टोकन की संख्या. एक टोकन करीब चार वर्णों का होता है. करीब 100 टोकन, करीब 60 से 80 शब्दों के होते हैं.

छोटे जवाबों के लिए कम वैल्यू और लंबे जवाबों के लिए ज़्यादा वैल्यू तय करें.

तापमान

तापमान का इस्तेमाल, जवाब देने के दौरान सैंपलिंग के लिए किया जाता है. यह तब होता है, जब topP और topK लागू किए जाते हैं. टोकन चुनने में रैंडमनेस की डिग्री को तापमान कंट्रोल करता है. कम तापमान ऐसे संकेतों के लिए अच्छा होता है जिनके लिए ज़्यादा तय और आसान जवाब की ज़रूरत होती है. वहीं, ज़्यादा तापमान होने से क्रिएटिव या अलग-अलग नतीजे हो सकते हैं. 0 का तापमान तय होता है. इसका मतलब है कि सबसे ज़्यादा संभावना वाले रिस्पॉन्स को हमेशा चुना जाता है.

इस्तेमाल के ज़्यादातर उदाहरणों के लिए, तापमान 0.2 से शुरू करें. अगर मॉडल बहुत सामान्य, बहुत छोटा जवाब देता है या मॉडल फ़ॉलबैक रिस्पॉन्स देता है, तो तापमान बढ़ाएं.

टॉप-के

Top-K, आउटपुट के लिए टोकन चुनने का तरीका बदलता है. 1 में टॉप-K का मतलब है कि मॉडल की शब्दावली में सभी टोकन में से, चुना गया अगला टोकन सबसे ज़्यादा है. इसे ग्रीडी डिकोडिंग भी कहा जाता है. वहीं, 3 का टॉप-K का मतलब है कि तापमान का इस्तेमाल करके, अगले टोकन को तीन सबसे सही टोकन में से चुना गया है.

टोकन चुनने के हर चरण के लिए, सबसे ज़्यादा संभावना वाले टॉप-K टोकन का सैंपल लिया जाता है. इसके बाद, टॉप-पी के आधार पर टोकन को और फ़िल्टर किया जाता है. तापमान के सैंपल का इस्तेमाल करके, आखिरी टोकन को चुना जाता है.

रैंडम रिस्पॉन्स के लिए कम वैल्यू और ज़्यादा रैंडम रिस्पॉन्स के लिए ज़्यादा वैल्यू तय करें. डिफ़ॉल्ट टॉप-K 40 है.

टॉप-पी

टॉप-P, आउटपुट के लिए टोकन चुनने का तरीका बदलता है. टोकन सबसे ज़्यादा (टॉप-K देखें) से लेकर सबसे कम संभावना तक चुने जाते हैं. ऐसा तब तक किया जाता है, जब तक कि उनकी संभावनाओं का योग, टॉप-P वैल्यू के बराबर न हो. उदाहरण के लिए, अगर टोकन A, B, और C की संभावना 0.3, 0.2, और 0.1 है और टॉप-P वैल्यू 0.5 है, तो मॉडल तापमान का इस्तेमाल करके अगले टोकन के तौर पर A या B चुनेगा और C को कैंडिडेट के तौर पर शामिल नहीं करेगा.

रैंडम रिस्पॉन्स के लिए कम वैल्यू और ज़्यादा रैंडम रिस्पॉन्स के लिए ज़्यादा वैल्यू तय करें. डिफ़ॉल्ट टॉप-P, 0.95 है.

प्रॉम्प्ट दोहराए जाने की रणनीतियां

प्रॉम्प्ट डिज़ाइन, बार-बार इस्तेमाल की जाने वाली एक प्रोसेस है. इसमें अपनी पसंद का जवाब पाने से पहले, अक्सर इसे कई बार दोहराना पड़ता है. इस सेक्शन में कुछ ऐसी चीज़ों के बारे में जानकारी दी गई है जिन्हें अपने प्रॉम्प्ट पर बार बार लागू करते समय आज़माया जा सकता है.

अलग-अलग वाक्यांशों का इस्तेमाल करें

अपने प्रॉम्प्ट में अलग-अलग शब्दों या वाक्यांश का इस्तेमाल करने से, मॉडल से अलग-अलग रिस्पॉन्स मिलते हैं. भले ही, उन सभी का मतलब एक ही हो. अगर आपको अपने प्रॉम्प्ट से उम्मीद के मुताबिक नतीजे नहीं मिल रहे हैं, तो इसे अलग तरीके से बोलकर देखें.

Version 1:
How do I bake a pie?

Version 2:
Suggest a recipe for a pie.

Version 3:
What's a good pie recipe?
  

मिलते-जुलते टास्क पर स्विच करें

अगर आपको टास्क के लिए दिए गए निर्देशों का पालन करने के लिए मॉडल नहीं मिल पा रहा है, तो इसे किसी ऐसे टास्क के लिए निर्देश दें जिससे वही नतीजा मिलता हो.

यह प्रॉम्प्ट, मॉडल को किताब की कैटगरी तय करने के लिए कहता है. इसके लिए, पहले से तय कैटगरी का इस्तेमाल किया जाता है.

Which category does The Odyssey belong to:
thriller
sci-fi
mythology
biography
  
The answer: mythology

Explanations:

The Odyssey is an epic poem in ancient Greek literature. It is one of two major ancient Greek epic
poems attributed to Homer. The Odyssey is an epic poem that tells the story of Odysseus, a Greek
hero who fought in the Trojan War and then spent ten years trying to return home to Ithaca. The
Odyssey is a work of mythology.
  

जवाब सही है, लेकिन मॉडल विकल्पों की सीमाओं के अंदर नहीं रहा. आप चाहें, तो पूरे वाक्य में जवाब देने के बजाय, सिर्फ़ किसी एक विकल्प का इस्तेमाल करके जवाब दिया जाना चाहिए. इस मामले में, निर्देशों को कई विकल्प वाले सवाल के तौर पर अलग तरीके से लिखा जा सकता है. साथ ही, मॉडल को कोई विकल्प चुनने के लिए कहा जा सकता है.

Multiple choice problem: Which of the following options describes the book The Odyssey?
Options:
- thriller
- sci-fi
- mythology
- biography
  
The answer is: mythology
  

प्रॉम्प्ट वाले कॉन्टेंट का क्रम बदलना

प्रॉम्प्ट में कॉन्टेंट के क्रम की वजह से, कभी-कभी जवाब पर असर पड़ सकता है. कॉन्टेंट के क्रम को बदलकर देखें और देखें कि इससे रिस्पॉन्स पर क्या असर पड़ता है.

Version 1:
[examples]
[context]
[input]

Version 2:
[input]
[examples]
[context]

Version 3:
[examples]
[input]
[context]

फ़ॉलबैक रिस्पॉन्स

फ़ॉलबैक रिस्पॉन्स, एक ऐसा रिस्पॉन्स है जो मॉडल तब भेजता है, जब प्रॉम्प्ट या जवाब किसी सुरक्षा फ़िल्टर को ट्रिगर करता है. फ़ॉलबैक रिस्पॉन्स का एक उदाहरण यह है, "मैं इसमें कोई मदद नहीं कर सकता, क्योंकि मैं सिर्फ़ एक लैंग्वेज मॉडल हूं."

अगर मॉडल, फ़ॉलबैक रिस्पॉन्स के साथ जवाब देता है, तो तापमान बढ़ाएं.

ऐसा करने से बचें

  • तथ्यों के साथ जानकारी देने के लिए मॉडल के भरोसे रहने से बचें.
  • गणित और तर्क से जुड़े सवालों को ध्यान में रखें.

अगले चरण

  • प्रॉम्प्ट के डिज़ाइन को अच्छी तरह से समझने के बाद, Google AI Studio का इस्तेमाल करके अपने प्रॉम्प्ट लिखें.
  • मल्टीमोडल प्रॉम्प्ट के बारे में जानने के लिए, मीडिया फ़ाइलों से अनुरोध करना लेख देखें.