Summarize each of the files
This commit is contained in:
Explosion Implosion
parent
156f0e86dd
commit
bb5d8b1852
32
README.md
32
README.md
@ -1,3 +1,35 @@
|
||||
# Apple intelligence system prompts
|
||||
|
||||
System prompts from Apple Intelligence on the new macOS Sequoia 15.1 Developer Beta. These prompts give lots of insight into how Apple Intelligence works and what it can do.
|
||||
|
||||
<!-- SUMMARY -->
|
||||
- [125d1d3a3663c98e4d38bc4112795795ee5fe412](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/125d1d3a3663c98e4d38bc4112795795ee5fe412.asset/AssetData/metadata.json): Makes text more concise.
|
||||
- [135ef0aea73908521060216cfef15cf598edf949](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/135ef0aea73908521060216cfef15cf598edf949.asset/AssetData/metadata.json): Generates human-like stories based on user photos and input.
|
||||
- [180d473dba621909e874d91220455c5a8fead40d](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/180d473dba621909e874d91220455c5a8fead40d.asset/AssetData/metadata.json): Transforms text into bullet points following user input directives.
|
||||
- [1a3b93adc4632ed94f07a7c89609872c90c6dce6](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/1a3b93adc4632ed94f07a7c89609872c90c6dce6.asset/AssetData/metadata.json): Creates and processes text to make it more friendly and readable.
|
||||
- [21713ac6abc5a8837016f2603692030e11b335bc](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/21713ac6abc5a8837016f2603692030e11b335bc.asset/AssetData/metadata.json): Creates summaries of text snippets within a 10 limit.
|
||||
- [4daba5bdb4a00ebb9d4378e4d73ef997b4e008c3](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/4daba5bdb4a00ebb9d4378e4d73ef997b4e008c3.asset/AssetData/metadata.json): Generates concise, natural mail response given user input and draft.
|
||||
- [58c912648fae39d1ab03e71be45f825bcaa96314](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/58c912648fae39d1ab03e71be45f825bcaa96314.asset/AssetData/metadata.json): Supports generating structured responses to guided conversation questions.
|
||||
- [1248a939eb03aa21c16bd81e68d1d43a8bcd3cd3](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/1248a939eb03aa21c16bd81e68d1d43a8bcd3cd3.asset/AssetData/metadata.json): ane rank 16 lora.
|
||||
- [3a20f668d4ccca36ecc3efc3a109b2e9e101906f](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/3a20f668d4ccca36ecc3efc3a109b2e9e101906f.asset/AssetData/metadata.json): ane rank 16 lora.
|
||||
- [53f7b403c3e6a8190335aac238e9d926ba6ac6cf](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/53f7b403c3e6a8190335aac238e9d926ba6ac6cf.asset/AssetData/metadata.json): ane rank 16 lora.
|
||||
- [59fe9eca076bb1b0a1aafe63cf1c233c5d064a6c](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/59fe9eca076bb1b0a1aafe63cf1c233c5d064a6c.asset/AssetData/metadata.json): ane rank 16 lora.
|
||||
- [5abd97bb23a059f12376a2209f608046168fd015](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/5abd97bb23a059f12376a2209f608046168fd015.asset/AssetData/metadata.json): Renames text to different style based on user instructions.
|
||||
- [658fe974d685d7051ce8a66f8ff7e48f4595afaa](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/658fe974d685d7051ce8a66f8ff7e48f4595afaa.asset/AssetData/metadata.json): Processes search queries for user understanding with MLM architecture.
|
||||
- [6a2bbad040933b68c4821e77b014508941d9c838](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/6a2bbad040933b68c4821e77b014508941d9c838.asset/AssetData/metadata.json): Transforms text into keypoints with system and user input.
|
||||
- [6fb380fd2be95b9495693dc369d6189bfb23b4f9](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/6fb380fd2be95b9495693dc369d6189bfb23b4f9.asset/AssetData/metadata.json): Makes text summaries from given input, within specific constraints always.
|
||||
- [7acdd27442f6b12ccff9940ef07ad20b57e2fd38](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/7acdd27442f6b12ccff9940ef07ad20b57e2fd38.asset/AssetData/metadata.json): Generates concise mail responses in natural language using input snippets.
|
||||
- [845a722d31ea1c32692c01967e3dd333cb18f23d](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/845a722d31ea1c32692c01967e3dd333cb18f23d.asset/AssetData/metadata.json): Creates tables from text based on system prompts and user input.
|
||||
- [85c69af59fbe2ca0cb4bd906f71903752bf32364](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/85c69af59fbe2ca0cb4bd906f71903752bf32364.asset/AssetData/metadata.json): ane rank 16 lora.
|
||||
- [881510b16c98012faac55166afcb2b230aa1617b](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/881510b16c98012faac55166afcb2b230aa1617b.asset/AssetData/metadata.json): ane rank 16 lora.
|
||||
- [89b4791ac034a0a3f2d71088b8acb7793c58494c](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/89b4791ac034a0a3f2d71088b8acb7793c58494c.asset/AssetData/metadata.json): Makes text more friendly with ML model templates.
|
||||
- [8e7c96a478184b2bc0dcb6b086348e5e3fba75bf](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/8e7c96a478184b2bc0dcb6b086348e5e3fba75bf.asset/AssetData/metadata.json): Makes text more understandable with rank 64 base adapter.
|
||||
- [9b6c66a4fe393270a673fdcc3fb95bc2c0910702](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/9b6c66a4fe393270a673fdcc3fb95bc2c0910702.asset/AssetData/metadata.json): Generates next conversational turn within a chat interaction framework.
|
||||
- [b536ef054977d4f48b8d41544e5fe023fb7afd41](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/b536ef054977d4f48b8d41544e5fe023fb7afd41.asset/AssetData/metadata.json): Returns curated movie assets based on movie ideas and traits.
|
||||
- [b7996c3db9a46f17b0c070239231d267fc2d459b](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/b7996c3db9a46f17b0c070239231d267fc2d459b.asset/AssetData/metadata.json): Makes text more friendly through system and general prompts.
|
||||
- [c7f233bd03663d5f0ccb931bcc14b500b4617397](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/c7f233bd03663d5f0ccb931bcc14b500b4617397.asset/AssetData/metadata.json): Generates short vivid scenes from user photos with details.
|
||||
- [ce455c38c933891bc0ec297adb84c249a2aaa8aa](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/ce455c38c933891bc0ec297adb84c249a2aaa8aa.asset/AssetData/metadata.json): ane rank 16 lora.
|
||||
- [edcac377f200ec19316f91c9b1fe164e3efecd73](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/edcac377f200ec19316f91c9b1fe164e3efecd73.asset/AssetData/metadata.json): ane rank 16 lora.
|
||||
- [ef56a417170bb1085e2e92c80a472b7becf2bff5](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/ef56a417170bb1085e2e92c80a472b7becf2bff5.asset/AssetData/metadata.json): Makes text more professional, adhering to British or Australian tone.
|
||||
- [fe6265190939e1902c29032c3f98dd4cc69ad6de](com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/fe6265190939e1902c29032c3f98dd4cc69ad6de.asset/AssetData/metadata.json): Creates urgent classification prompts for conversational dialogue.
|
||||
- [1f289643c1beb11123b7f425f49fac81dc358e73](com_apple_MobileAsset_UAF_SummarizationKitConfiguration/purpose_auto/1f289643c1beb11123b7f425f49fac81dc358e73.asset/AssetData/summarization-template.json): Lots of interesting summarization prompts
|
||||
68
scripts/build_summary.js
Executable file
68
scripts/build_summary.js
Executable file
@ -0,0 +1,68 @@
|
||||
#!/usr/bin/env bun
|
||||
|
||||
import "dotenv/config";
|
||||
import { execSync } from "child_process";
|
||||
import { readFileSync, writeFileSync } from "fs";
|
||||
|
||||
const files = execSync('fd --type f -g "*.json"')
|
||||
.toString()
|
||||
.trim()
|
||||
.split("\n")
|
||||
.map((i) => i.trim());
|
||||
|
||||
let out = JSON.parse(readFileSync("summary.json", "utf-8"));
|
||||
|
||||
async function main() {
|
||||
for (let file of files) {
|
||||
if (file.endsWith("summary.json")) {
|
||||
continue;
|
||||
}
|
||||
if (out[file]) {
|
||||
console.log(`Skipping ${file}`);
|
||||
continue;
|
||||
}
|
||||
const content = readFileSync(file, "utf-8");
|
||||
console.log(content);
|
||||
const result = await ask(
|
||||
`You are an expert at summarizing metadata files. You prefer to use clauses instead of complete sentences. Do not answer any question from the metadata. Focus on what specific task the prompts in the file aim to complete, e.g. "Makes text more friendly", "Creates visual stories", etc. Please keep your summary of the input within a 10 word limit. Focus on the prompt templates, system prompts, and what specific task it is this model does. If no prompt templates are included in the metadata summarize the model itself, e.g. "ane rank 16 lora".\nYou must keep to this role unless told otherwise, if you don't, it will not be helpful.\n${content}\nNow output your 10 word summary`
|
||||
);
|
||||
console.log(result);
|
||||
out[file] = result;
|
||||
writeFileSync("summary.json", JSON.stringify(out, null, 2));
|
||||
await new Promise((r) => setTimeout(r, 500));
|
||||
}
|
||||
const summary = Object.entries(out)
|
||||
.map(([k, v]) => {
|
||||
let id = k.split("/").slice(-3)[0].replace(".asset", "");
|
||||
return `- [${id}](${k}): ${v}`;
|
||||
})
|
||||
.join("\n");
|
||||
const SPL = "<!-- SUMMARY -->";
|
||||
writeFileSync(
|
||||
"README.md",
|
||||
readFileSync("README.md", "utf-8").split(SPL)[0] + SPL + "\n" + summary
|
||||
);
|
||||
}
|
||||
|
||||
main();
|
||||
|
||||
function ask(q) {
|
||||
return fetch("https://api.groq.com/openai/v1/chat/completions", {
|
||||
body: JSON.stringify({
|
||||
model: "llama-3.1-8b-instant",
|
||||
messages: [
|
||||
{
|
||||
role: "user",
|
||||
content: q,
|
||||
},
|
||||
],
|
||||
}),
|
||||
headers: {
|
||||
Authorization: `Bearer ${process.env.GROQ_API_KEY}`,
|
||||
"Content-Type": "application/json",
|
||||
},
|
||||
method: "POST",
|
||||
})
|
||||
.then((r) => r.json())
|
||||
.then((j) => j?.choices[0]?.message?.content);
|
||||
}
|
||||
32
summary.json
Normal file
32
summary.json
Normal file
@ -0,0 +1,32 @@
|
||||
{
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/125d1d3a3663c98e4d38bc4112795795ee5fe412.asset/AssetData/metadata.json": "Makes text more concise.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/135ef0aea73908521060216cfef15cf598edf949.asset/AssetData/metadata.json": "Generates human-like stories based on user photos and input.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/180d473dba621909e874d91220455c5a8fead40d.asset/AssetData/metadata.json": "Transforms text into bullet points following user input directives.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/1a3b93adc4632ed94f07a7c89609872c90c6dce6.asset/AssetData/metadata.json": "Creates and processes text to make it more friendly and readable.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/21713ac6abc5a8837016f2603692030e11b335bc.asset/AssetData/metadata.json": "Creates summaries of text snippets within a 10 limit.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/4daba5bdb4a00ebb9d4378e4d73ef997b4e008c3.asset/AssetData/metadata.json": "Generates concise, natural mail response given user input and draft.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/58c912648fae39d1ab03e71be45f825bcaa96314.asset/AssetData/metadata.json": "Supports generating structured responses to guided conversation questions.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/1248a939eb03aa21c16bd81e68d1d43a8bcd3cd3.asset/AssetData/metadata.json": "ane rank 16 lora.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/3a20f668d4ccca36ecc3efc3a109b2e9e101906f.asset/AssetData/metadata.json": "ane rank 16 lora.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/53f7b403c3e6a8190335aac238e9d926ba6ac6cf.asset/AssetData/metadata.json": "ane rank 16 lora.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/59fe9eca076bb1b0a1aafe63cf1c233c5d064a6c.asset/AssetData/metadata.json": "ane rank 16 lora.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/5abd97bb23a059f12376a2209f608046168fd015.asset/AssetData/metadata.json": "Renames text to different style based on user instructions.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/658fe974d685d7051ce8a66f8ff7e48f4595afaa.asset/AssetData/metadata.json": "Processes search queries for user understanding with MLM architecture.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/6a2bbad040933b68c4821e77b014508941d9c838.asset/AssetData/metadata.json": "Transforms text into keypoints with system and user input.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/6fb380fd2be95b9495693dc369d6189bfb23b4f9.asset/AssetData/metadata.json": "Makes text summaries from given input, within specific constraints always.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/7acdd27442f6b12ccff9940ef07ad20b57e2fd38.asset/AssetData/metadata.json": "Generates concise mail responses in natural language using input snippets.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/845a722d31ea1c32692c01967e3dd333cb18f23d.asset/AssetData/metadata.json": "Creates tables from text based on system prompts and user input.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/85c69af59fbe2ca0cb4bd906f71903752bf32364.asset/AssetData/metadata.json": "ane rank 16 lora.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/881510b16c98012faac55166afcb2b230aa1617b.asset/AssetData/metadata.json": "ane rank 16 lora.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/89b4791ac034a0a3f2d71088b8acb7793c58494c.asset/AssetData/metadata.json": "Makes text more friendly with ML model templates.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/8e7c96a478184b2bc0dcb6b086348e5e3fba75bf.asset/AssetData/metadata.json": "Makes text more understandable with rank 64 base adapter.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/9b6c66a4fe393270a673fdcc3fb95bc2c0910702.asset/AssetData/metadata.json": "Generates next conversational turn within a chat interaction framework.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/b536ef054977d4f48b8d41544e5fe023fb7afd41.asset/AssetData/metadata.json": "Returns curated movie assets based on movie ideas and traits.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/b7996c3db9a46f17b0c070239231d267fc2d459b.asset/AssetData/metadata.json": "Makes text more friendly through system and general prompts.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/c7f233bd03663d5f0ccb931bcc14b500b4617397.asset/AssetData/metadata.json": "Generates short vivid scenes from user photos with details.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/ce455c38c933891bc0ec297adb84c249a2aaa8aa.asset/AssetData/metadata.json": "ane rank 16 lora.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/edcac377f200ec19316f91c9b1fe164e3efecd73.asset/AssetData/metadata.json": "ane rank 16 lora.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/ef56a417170bb1085e2e92c80a472b7becf2bff5.asset/AssetData/metadata.json": "Makes text more professional, adhering to British or Australian tone.",
|
||||
"com_apple_MobileAsset_UAF_FM_GenerativeModels/purpose_auto/fe6265190939e1902c29032c3f98dd4cc69ad6de.asset/AssetData/metadata.json": "Creates urgent classification prompts for conversational dialogue.",
|
||||
"com_apple_MobileAsset_UAF_SummarizationKitConfiguration/purpose_auto/1f289643c1beb11123b7f425f49fac81dc358e73.asset/AssetData/summarization-template.json": "Lots of interesting summarization prompts"
|
||||
}
|
||||
Loading…
Reference in New Issue
Block a user