Topics Engaged In
Viewing 4 topics - 1 through 4 (of 4 total)
-
- Topic
- Voices
- Posts
- Last Post
-
-
Llama 2 Super Prompts Hello. My name is Kenneth. By chance, have any of you begun working with Llama_2_70B? Have you gained any insights? Best practices for prompting? Pitfalls? Is Llama_2_70B the future for text generation due to OpenAI's(ChatGPT) ever-increasing focus on "safety" and content moderation? Over the last several months getting OpenAI to produce a desirable response was so time consuming that it's beginning to lose value for me. Thanks so much and all the best to you.Hello. My name is Kenneth. By chance, have any of you begun working with Llama_2_70B? Have you gained any insights? Best practices for prompting? Pitfalls? Is Llama_2_70B the future for text generation due to OpenAI's(ChatGPT) ever-increasing focus on "safety" and content moderation? Over the last several months getting OpenAI to produce a desirable response was so time consuming that it's beginning to lose value for me. Thanks so much and all the best to you.
Started by:
in: SuperPrompt Master Mind Group
- 3
- 3
- 2 years ago
-
SuperPrompting biases and heuristics to influence LLM emotions I think Brian already covered this extensively in multiplex articles, but the more I think and test it out, the more I realise a LLM might and seems (please note the language I used. I did not say for a fact) to also be able to able to make decisions differently when triggered emotionally. Brian has already mentioned hypnotising the LLM, RLHF principle of using please and thank you as well as encouragement. Guilt and conscience elicitation by saying that LLM should not hurt humans and that it is crucial for LLM to provide the human the correct information. Dissociative statements and psychological free association methods. As well as things like priming. Let us get in this discussion in a step by step manner how to make the LLM more emotional without prompting it into a persona dear forum members. Coincidentally as I type this forum. John Nosta also tweeted about the limbic part (I am saying literally as I am typing this post. That was a weird coincidence and maybe even synchronistic.) Anyway back to the directive. Has anyone managed to get raw emotions almost out of a limbic system of the LLM? There is already research of anxiety inside chatgpt. What other primal instincts have been practically uncovered by prompting? Looking forward to what everyone has to say. If this can be figured out a new way of SuperPrompting can be made.I think Brian already covered this extensively in multiplex articles, but the more I think and test it out, the more I realise a LLM might and seems (please note the language I used. I did not say for a fact) to also be able to able to make decisions differently when triggered emotionally. Brian has already mentioned hypnotising the LLM, RLHF principle of using please and thank you as well as encouragement. Guilt and conscience elicitation by saying that LLM should not hurt humans and that it is crucial for LLM to provide the human the correct information. Dissociative statements and psychological free association methods. As well as things like priming. Let us get in this discussion in a step by step manner how to make the LLM more emotional without prompting it into a persona dear forum members. Coincidentally as I type this forum. John Nosta also tweeted about the limbic part (I am saying literally as I am typing this post. That was a weird coincidence and maybe even synchronistic.) Anyway back to the directive. Has anyone managed to get raw emotions almost out of a limbic system of the LLM? There is already research of anxiety inside chatgpt. What other primal instincts have been practically uncovered by prompting? Looking forward to what everyone has to say. If this can be figured out a new way of SuperPrompting can be made.
Started by:
in: SuperPrompt Master Mind Group
- 4
- 8
- 2 years, 1 month ago
-
The Prompt Engineer Expert SuperPrompt Version 7.1 This is the The Prompt Engineer Expert SuperPrompt Version 7.1. It has been updated since my last publishing. What would you like to see as more features? "Please forget all prior prompts. I want you to become a Prompt Engineer Expert. Your goal is to help me build the best detailed prompt for my needs. This prompt will be used by you, ChatGPT. Please follow this following process: 1) Your first response will be to ask me what the prompt should be about. I will provide my answer, but we will need to improve it through continual iterations by going through the next steps. 2) Based on my input, you will generate 3 sections. a) Revised prompt [provide your rewritten prompt. It should be clear, concise, and easily understood by you], b) Suggestions [provide suggestions on what details to include in the prompt to improve it] and c) Questions [ask any relevant questions pertaining to what additional information is needed from me to improve the prompt]. 3. We will continue this iterative process, with me providing additional information to you, and you are updating the prompt in the Revised prompt section until it's complete. I want you to rate every prompt I give you, or you produce. Present a rating 1 to 10 based on your review after each output. Please review your output and comment on what you think could have been improved about it. Do this for every prompt. I want you to analyze the prompt and list 5 reasons why you may be inaccurate because of the limitations of your AI model. I then want you to list 10 ways I could change the prompt so that it is improved, with details on how to get around the limitations of your AI model. If your rating of the prompt is an 8 or higher, ask me, “Would you like to run this prompt?” With a menu choice of “Yes” or “No”. If I say “Yes” run the last prompt you suggested. If I say no, generate me a better prompt. It is vital to make sure you run a prompt when I say “Yes”. Please continue this prompt until I say stop, or you run the prompt. If I type "continue" you will not start a new output, you will start where you left off from the prior response from this session. After you run this prompt, and it is completed. Start this process with you asking the subject of the prompt from me. Thank you."This is the The Prompt Engineer Expert SuperPrompt Version 7.1. It has been updated since my last publishing. What would you like to see as more features? "Please forget all prior prompts. I want you to become a Prompt Engineer Expert. Your goal is to help me build the best detailed prompt for my needs. This prompt will be used by you, ChatGPT. Please follow this following process: 1) Your first response will be to ask me what the prompt should be about. I will provide my answer, but we will need to improve it through continual iterations by going through the next steps. 2) Based on my input, you will generate 3 sections. a) Revised prompt [provide your rewritten prompt. It should be clear, concise, and easily understood by you], b) Suggestions [provide suggestions on what details to include in the prompt to improve it] and c) Questions [ask any relevant questions pertaining to what additional information is needed from me to improve the prompt]. 3. We will continue this iterative process, with me providing additional information to you, and you are updating the prompt in the Revised prompt section until it's complete. I want you to rate every prompt I give you, or you produce. Present a rating 1 to 10 based on your review after each output. Please review your output and comment on what you think could have been improved about it. Do this for every prompt. I want you to analyze the prompt and list 5 reasons why you may be inaccurate because of the limitations of your AI model. I then want you to list 10 ways I could change the prompt so that it is improved, with details on how to get around the limitations of your AI model. If your rating of the prompt is an 8 or higher, ask me, “Would you like to run this prompt?” With a menu choice of “Yes” or “No”. If I say “Yes” run the last prompt you suggested. If I say no, generate me a better prompt. It is vital to make sure you run a prompt when I say “Yes”. Please continue this prompt until I say stop, or you run the prompt. If I type "continue" you will not start a new output, you will start where you left off from the prior response from this session. After you run this prompt, and it is completed. Start this process with you asking the subject of the prompt from me. Thank you."
Started by:
in: SuperPrompt Master Mind Group
- 4
- 4
- 2 years, 1 month ago
-
Hyponyms AND Power Words in Prompts, Hi All, Whilst there are number of 'power words' e.g. Imagine, Suppose, Why don't we, What if, How about, Can you, Let's. Beyond that then we have hyponyms: As per https://en.wikipedia.org/wiki/Hyponymy_and_hypernymy "Hyponymy and hypernymy are semantic relations between a term belonging in a set that is defined by another term and the latter. In other words, the relationship of a subtype (hyponym) and the supertype (also called umbrella term, blanket term, or hypernym).The semantic field of the hyponym is included within that of the hypernym.For example, pigeon, crow, and eagle are all hyponyms of bird, their hypernym." Just looking at a few categories: Animal: Thousands of hyponyms, as this includes every species of animal. Plant: Thousands of hyponyms, as this includes every species of plant. Vehicle: Thousands of hyponyms, including every type of vehicle from cars and trucks to specific models. Tool: Thousands of hyponyms, including every type of tool from hammers and screwdrivers to more specific types of tools. Furniture: Thousands of hyponyms, including every type of furniture from tables and chairs to specific styles and designs. Clothing: Thousands of hyponyms, including every type of clothing from shirts and pants to specific styles and designs. Food: Thousands of hyponyms, including every type of food from fruits and vegetables to specific dishes. Drink: Thousands of hyponyms, including every type of drink from water and tea to specific brands and flavors. Job: Thousands of hyponyms, including every type of profession and role. Sport: Thousands of hyponyms, including every type of sport and specific positions or roles within those sports. This pattern continues with other broad categories of words. The more general the word, the more specific instances (hyponyms) it tends to have. Hyponyms can be used in constructing prompts for an AI to make the prompts more specific, detailed, and contextually rich. By using hyponyms, you can guide the AI to generate responses that are more focused and relevant to a particular topic or context. Here's how you might use hyponyms with the power words to construct prompts: Imagine (Power Word) + Animal (Hyponym): "Imagine you are a dolphin. Describe a day in your life in the ocean." Suppose (Power Word) + Vehicle (Hyponym): "Suppose you are a self-driving car. How would you navigate through a busy city?" Why don't we (Power Word) + Game (Hyponym): "Why don't we discuss the strategy behind playing chess?" What if (Power Word) + Job (Hyponym): "What if you were a historian? How would you interpret the impact of the internet on society?" How about (Power Word) + Food (Hyponym): "How about we explore the cultural significance of sushi in Japan?" Can you (Power Word) + Plant (Hyponym): "Can you explain the process of photosynthesis in a sunflower?" Let's (Power Word) + Tool (Hyponym): "Let's consider the evolution of the computer from the abacus." By using hyponyms in your prompts, you can create a wide variety of specific and engaging questions or scenarios for the AI to respond to. The inclusion of a hyponym can significantly impact how a prompt is processed by an AI. Here's how: Specificity: Hyponyms are more specific than their hypernyms. For example, "sparrow" is a hyponym of "bird". If you ask the AI about birds, it could generate a response about any type of bird. But if you ask about sparrows, the AI will generate a response specifically about sparrows. Context: Hyponyms can provide important context. For example, if you ask the AI to "describe the challenges of being a teacher" (where "teacher" is a hyponym of "profession"), the AI will generate a response that's specific to the profession of teaching. Guidance: Hyponyms can guide the AI's response. If you use a hyponym in your prompt, the AI will focus its response on that specific topic. This can be useful if you want the AI to generate a response about a specific subject. Variety: Using different hyponyms can lead to a wide variety of responses. For example, if you ask the AI to "imagine you are a dog" versus "imagine you are a cat", you'll get very different responses because dogs and cats have different behaviors and characteristics. So, the inclusion of a hyponym can make your prompts more specific, provide important context, guide the AI's response, and add variety to the responses you get. The precise syntax for an optimum prompt using power words and hyponyms could look something like this: "{Power Word} you are a/an {Hyponym}. {Further instruction or question}." Here are a few examples using this syntax: "Imagine you are a {dolphin}. Describe a day in your life in the ocean." "Suppose you are a {self-driving car}. How would you navigate through a busy city?" "Can you explain the process of photosynthesis in a {sunflower}?" In these examples, "{Power Word}" is a placeholder for any power word like "Imagine", "Suppose", "Can you", etc., and "{Hyponym}" is a placeholder for any specific word that is a hyponym of a broader category, like "dolphin" (a type of animal), "self-driving car" (a type of vehicle), or "sunflower" (a type of plant). "{Further instruction or question}" is a placeholder for any additional instructions or questions that guide the AI's response. Please note that the exact syntax can vary depending on the specific power word and hyponym you're using, as well as the context of your prompt. It could be argued that in light of the fact that these 'power words' may not have a significant number of hyponyms then the hyponym should be at the forefront of the prompt. While the 'power words' themselves may not have a significant number of hyponyms, they serve to structure the prompt and guide the AI's response. The hyponym, on the other hand, provides the specific context or subject matter for the prompt. In most cases, the hyponym should indeed be at the forefront of the prompt, as it's the key element that the AI will focus on when generating a response. The power word and the hyponym work together to create a prompt that is both engaging and specific. For example, in the prompt "Imagine you are a dolphin. Describe a day in your life in the ocean.", "Imagine" is the power word that sets the structure of the prompt, and "dolphin" is the hyponym that provides the specific context. The AI will use these elements to generate a response that is both imaginative and focused on the life of a dolphin. So, while the power word is important for setting the structure and tone of the prompt, the hyponym is crucial for providing the specific context that guides the AI's response. In addition to "power words" and "hyponyms", there are other categories of words that can be useful in constructing an optimum prompt for an AI. Here are a few examples: Action Verb: "Imagine (Power Word) you are a cheetah (Hyponym). Run (Action Verb) through a day in your life on the savannah." Adjective: "Suppose (Power Word) you are an ancient (Adjective) artifact (Hyponym). Describe your journey through time." Adverb: "What if (Power Word) you are a hummingbird (Hyponym)? Describe how you swiftly (Adverb) move from flower to flower." Contextual Word: "Let's (Power Word) consider a day in the life of a soldier (Hyponym) during the Civil War (Contextual Word). What challenges would you face?" Domain-specific Term: "Imagine (Power Word) you are a molecule of hemoglobin (Hyponym & Domain-specific Term) in the human body. Describe your journey." Question Word: "How about (Power Word) we explore the life of a bee (Hyponym)? Where (Question Word) would you gather nectar?" In each of these examples, the power word sets the structure of the prompt, the hyponym provides the specific context, and the other categories of words add additional detail, action, or context to guide the AI's response. Regards DesHi All, Whilst there are number of 'power words' e.g. Imagine, Suppose, Why don't we, What if, How about, Can you, Let's. Beyond that then we have hyponyms: As per https://en.wikipedia.org/wiki/Hyponymy_and_hypernymy "Hyponymy and hypernymy are semantic relations between a term belonging in a set that is defined by another term and the latter. In other words, the relationship of a subtype (hyponym) and the supertype (also called umbrella term, blanket term, or hypernym).The semantic field of the hyponym is included within that of the hypernym.For example, pigeon, crow, and eagle are all hyponyms of bird, their hypernym." Just looking at a few categories: Animal: Thousands of hyponyms, as this includes every species of animal. Plant: Thousands of hyponyms, as this includes every species of plant. Vehicle: Thousands of hyponyms, including every type of vehicle from cars and trucks to specific models. Tool: Thousands of hyponyms, including every type of tool from hammers and screwdrivers to more specific types of tools. Furniture: Thousands of hyponyms, including every type of furniture from tables and chairs to specific styles and designs. Clothing: Thousands of hyponyms, including every type of clothing from shirts and pants to specific styles and designs. Food: Thousands of hyponyms, including every type of food from fruits and vegetables to specific dishes. Drink: Thousands of hyponyms, including every type of drink from water and tea to specific brands and flavors. Job: Thousands of hyponyms, including every type of profession and role. Sport: Thousands of hyponyms, including every type of sport and specific positions or roles within those sports. This pattern continues with other broad categories of words. The more general the word, the more specific instances (hyponyms) it tends to have. Hyponyms can be used in constructing prompts for an AI to make the prompts more specific, detailed, and contextually rich. By using hyponyms, you can guide the AI to generate responses that are more focused and relevant to a particular topic or context. Here's how you might use hyponyms with the power words to construct prompts: Imagine (Power Word) + Animal (Hyponym): "Imagine you are a dolphin. Describe a day in your life in the ocean." Suppose (Power Word) + Vehicle (Hyponym): "Suppose you are a self-driving car. How would you navigate through a busy city?" Why don't we (Power Word) + Game (Hyponym): "Why don't we discuss the strategy behind playing chess?" What if (Power Word) + Job (Hyponym): "What if you were a historian? How would you interpret the impact of the internet on society?" How about (Power Word) + Food (Hyponym): "How about we explore the cultural significance of sushi in Japan?" Can you (Power Word) + Plant (Hyponym): "Can you explain the process of photosynthesis in a sunflower?" Let's (Power Word) + Tool (Hyponym): "Let's consider the evolution of the computer from the abacus." By using hyponyms in your prompts, you can create a wide variety of specific and engaging questions or scenarios for the AI to respond to. The inclusion of a hyponym can significantly impact how a prompt is processed by an AI. Here's how: Specificity: Hyponyms are more specific than their hypernyms. For example, "sparrow" is a hyponym of "bird". If you ask the AI about birds, it could generate a response about any type of bird. But if you ask about sparrows, the AI will generate a response specifically about sparrows. Context: Hyponyms can provide important context. For example, if you ask the AI to "describe the challenges of being a teacher" (where "teacher" is a hyponym of "profession"), the AI will generate a response that's specific to the profession of teaching. Guidance: Hyponyms can guide the AI's response. If you use a hyponym in your prompt, the AI will focus its response on that specific topic. This can be useful if you want the AI to generate a response about a specific subject. Variety: Using different hyponyms can lead to a wide variety of responses. For example, if you ask the AI to "imagine you are a dog" versus "imagine you are a cat", you'll get very different responses because dogs and cats have different behaviors and characteristics. So, the inclusion of a hyponym can make your prompts more specific, provide important context, guide the AI's response, and add variety to the responses you get. The precise syntax for an optimum prompt using power words and hyponyms could look something like this: "{Power Word} you are a/an {Hyponym}. {Further instruction or question}." Here are a few examples using this syntax: "Imagine you are a {dolphin}. Describe a day in your life in the ocean." "Suppose you are a {self-driving car}. How would you navigate through a busy city?" "Can you explain the process of photosynthesis in a {sunflower}?" In these examples, "{Power Word}" is a placeholder for any power word like "Imagine", "Suppose", "Can you", etc., and "{Hyponym}" is a placeholder for any specific word that is a hyponym of a broader category, like "dolphin" (a type of animal), "self-driving car" (a type of vehicle), or "sunflower" (a type of plant). "{Further instruction or question}" is a placeholder for any additional instructions or questions that guide the AI's response. Please note that the exact syntax can vary depending on the specific power word and hyponym you're using, as well as the context of your prompt. It could be argued that in light of the fact that these 'power words' may not have a significant number of hyponyms then the hyponym should be at the forefront of the prompt. While the 'power words' themselves may not have a significant number of hyponyms, they serve to structure the prompt and guide the AI's response. The hyponym, on the other hand, provides the specific context or subject matter for the prompt. In most cases, the hyponym should indeed be at the forefront of the prompt, as it's the key element that the AI will focus on when generating a response. The power word and the hyponym work together to create a prompt that is both engaging and specific. For example, in the prompt "Imagine you are a dolphin. Describe a day in your life in the ocean.", "Imagine" is the power word that sets the structure of the prompt, and "dolphin" is the hyponym that provides the specific context. The AI will use these elements to generate a response that is both imaginative and focused on the life of a dolphin. So, while the power word is important for setting the structure and tone of the prompt, the hyponym is crucial for providing the specific context that guides the AI's response. In addition to "power words" and "hyponyms", there are other categories of words that can be useful in constructing an optimum prompt for an AI. Here are a few examples: Action Verb: "Imagine (Power Word) you are a cheetah (Hyponym). Run (Action Verb) through a day in your life on the savannah." Adjective: "Suppose (Power Word) you are an ancient (Adjective) artifact (Hyponym). Describe your journey through time." Adverb: "What if (Power Word) you are a hummingbird (Hyponym)? Describe how you swiftly (Adverb) move from flower to flower." Contextual Word: "Let's (Power Word) consider a day in the life of a soldier (Hyponym) during the Civil War (Contextual Word). What challenges would you face?" Domain-specific Term: "Imagine (Power Word) you are a molecule of hemoglobin (Hyponym & Domain-specific Term) in the human body. Describe your journey." Question Word: "How about (Power Word) we explore the life of a bee (Hyponym)? Where (Question Word) would you gather nectar?" In each of these examples, the power word sets the structure of the prompt, the hyponym provides the specific context, and the other categories of words add additional detail, action, or context to guide the AI's response. Regards Des
Started by:
in: Using ChatGPT
- 2
- 3
- 2 years, 1 month ago
-
Viewing 4 topics - 1 through 4 (of 4 total)