AaronPo/the-biergarten-app
1
0
Fork 0
mirror of https://github.com/aaronpo97/the-biergarten-app.git synced 2026-07-17 01:47:22 +00:00
Code Issues Packages Projects Releases Wiki Activity

format

Browse Source
This commit is contained in:
Aaron Po
2026-06-21 13:13:02 -04:00
parent 4de0ea6638
commit ad97b0ea6c
27 changed files with 313 additions and 298 deletions
Download Patch File Download Diff File Expand all files Collapse all files

42
tooling/pipeline/src/data_generation/llama/generate_brewery.cc
View File

@@ -83,8 +83,8 @@ BreweryResult LlamaGenerator::GenerateBrewery(
/**
* RETRY LOOP with validation and error correction
* Attempts to generate valid brewery data up to 3 times, with feedback-based
* refinement
* Attempts to generate valid brewery data up to 3 times, with
* feedback-based refinement
*/
constexpr int max_attempts = 3;
std::string raw;
@@ -103,7 +103,7 @@ BreweryResult LlamaGenerator::GenerateBrewery(
{.level = LogLevel::Debug,
.phase = PipelinePhase::BreweryAndBeerGeneration,
.message = std::format("LlamaGenerator: raw output (attempt {}): {}",
attempt + 1, raw)});
attempt + 1, raw)});
}
// Validate output: parse JSON and check required fields
@@ -119,8 +119,9 @@ BreweryResult LlamaGenerator::GenerateBrewery(
logger_->Log(
{.level = LogLevel::Info,
.phase = PipelinePhase::BreweryAndBeerGeneration,
.message = std::format("LlamaGenerator: successfully generated brewery data on attempt {}",
attempt + 1)});
.message = std::format("LlamaGenerator: successfully generated "
"brewery data on attempt {}",
attempt + 1)});
}
return brewery;
@@ -133,33 +134,38 @@ BreweryResult LlamaGenerator::GenerateBrewery(
logger_->Log(
{.level = LogLevel::Warn,
.phase = PipelinePhase::BreweryAndBeerGeneration,
.message =
std::format("LlamaGenerator: malformed brewery JSON (attempt {}): {}",
.message = std::format(
"LlamaGenerator: malformed brewery JSON (attempt {}): {}",
attempt + 1, *validation_error)});
}
// Update prompt with error details to guide LLM toward correct output.
user_prompt = std::format(
"Your previous response was invalid. Error: {}\nReturn the thought "
"process before the JSON if needed, then return ONLY valid JSON with "
"exactly these keys, in this exact order: {{\"name_en\": \"<English "
"process before the JSON if needed, then return ONLY valid JSON "
"with "
"exactly these keys, in this exact order: {{\"name_en\": "
"\"<English "
"brewery name>\", \"description_en\": \"<English single-paragraph "
"description>\", \"name_local\": \"<local-language brewery name>\", "
"description>\", \"name_local\": \"<local-language brewery "
"name>\", "
"\"description_local\": \"<local-language single-paragraph "
"description>\"}}.\nDo not include markdown, comments, extra keys, or "
"literal placeholder values.\n\nKeep the JSON strings concise enough "
"description>\"}}.\nDo not include markdown, comments, extra keys, "
"or "
"literal placeholder values.\n\nKeep the JSON strings concise "
"enough "
"to fit within the token budget.\n\n{}",
*validation_error, retry_location);
}
// All retry attempts exhausted: log failure and throw exception
if (logger_) {
logger_->Log(
{.level = LogLevel::Error,
.phase = PipelinePhase::BreweryAndBeerGeneration,
.message = std::format(
"LlamaGenerator: malformed brewery response after {} attempts: {}",
max_attempts, last_error.empty() ? raw : last_error)});
logger_->Log({.level = LogLevel::Error,
.phase = PipelinePhase::BreweryAndBeerGeneration,
.message = std::format(
"LlamaGenerator: malformed brewery "
"response after {} attempts: {}",
max_attempts, last_error.empty() ? raw : last_error)});
}
throw std::runtime_error("LlamaGenerator: malformed brewery response");
}

52
tooling/pipeline/src/data_generation/llama/generate_user.cc
View File

@@ -41,17 +41,17 @@ UserResult LlamaGenerator::GenerateUser(const EnrichedCity& city,
const std::string system_prompt = prompt_directory_->Load("USER_GENERATION");
std::string user_prompt = std::format(
"## NAME:\n{} {}\n\n## GENDER:\n{}\n\n## CITY:\n{}\n\n## COUNTRY:\n{}\n\n"
"## NAME:\n{} {}\n\n## GENDER:\n{}\n\n## CITY:\n{}\n\n## "
"COUNTRY:\n{}\n\n"
"## PERSONA:\n{}\n\n## PERSONA DESCRIPTION:\n{}\n\n## STYLE "
"AFFINITIES:\n{}",
name.first_name, name.last_name, name.gender, city.location.city,
city.location.country, persona.name, persona.description,
style_affinities);
const std::string retry_context =
std::format("Name: {} {}\nCity: {}, {}\nPersona: {}", name.first_name,
name.last_name, city.location.city, city.location.country,
persona.name);
const std::string retry_context = std::format(
"Name: {} {}\nCity: {}, {}\nPersona: {}", name.first_name, name.last_name,
city.location.city, city.location.country, persona.name);
constexpr int max_attempts = 3;
std::string raw;
@@ -59,14 +59,13 @@ UserResult LlamaGenerator::GenerateUser(const EnrichedCity& city,
int max_tokens = kUserInitialMaxTokens;
for (int attempt = 0; attempt < max_attempts; ++attempt) {
raw = this->Infer(system_prompt, user_prompt, max_tokens,
kUserJsonGrammar);
raw = this->Infer(system_prompt, user_prompt, max_tokens, kUserJsonGrammar);
if (logger_) {
logger_->Log(
{.level = LogLevel::Debug,
.phase = PipelinePhase::UserGeneration,
.message = std::format("LlamaGenerator: raw output (attempt {}): {}",
attempt + 1, raw)});
attempt + 1, raw)});
}
UserResult user;
@@ -78,8 +77,9 @@ UserResult LlamaGenerator::GenerateUser(const EnrichedCity& city,
logger_->Log(
{.level = LogLevel::Info,
.phase = PipelinePhase::UserGeneration,
.message = std::format("LlamaGenerator: successfully generated user data on attempt {}",
attempt + 1)});
.message = std::format("LlamaGenerator: successfully "
"generated user data on attempt {}",
attempt + 1)});
}
user.first_name = name.first_name;
@@ -90,32 +90,34 @@ UserResult LlamaGenerator::GenerateUser(const EnrichedCity& city,
last_error = *validation_error;
if (logger_) {
logger_->Log(
{.level = LogLevel::Warn,
.phase = PipelinePhase::UserGeneration,
.message =
std::format("LlamaGenerator: malformed user JSON (attempt {}): {}",
attempt + 1, *validation_error)});
logger_->Log({.level = LogLevel::Warn,
.phase = PipelinePhase::UserGeneration,
.message = std::format(
"LlamaGenerator: malformed user JSON (attempt {}): {}",
attempt + 1, *validation_error)});
}
user_prompt = std::format(
"Your previous response was invalid. Error: {}\nReturn the thought "
"process before the JSON if needed, then return ONLY valid JSON with "
"exactly these keys, in this exact order: {{\"username\": \"<handle "
"process before the JSON if needed, then return ONLY valid JSON "
"with "
"exactly these keys, in this exact order: {{\"username\": "
"\"<handle "
"derived from the given name>\", \"bio\": \"<first-person bio "
"grounded in the persona>\", \"activity_weight\": <number between 0 "
"grounded in the persona>\", \"activity_weight\": <number between "
"0 "
"and 1>}}.\nDo not include markdown, comments, extra keys, or "
"literal placeholder values.\n\n{}",
*validation_error, retry_context);
}
if (logger_) {
logger_->Log(
{.level = LogLevel::Error,
.phase = PipelinePhase::UserGeneration,
.message = std::format(
"LlamaGenerator: malformed user response after {} attempts: {}",
max_attempts, last_error.empty() ? raw : last_error)});
logger_->Log({.level = LogLevel::Error,
.phase = PipelinePhase::UserGeneration,
.message = std::format(
"LlamaGenerator: malformed user response "
"after {} attempts: {}",
max_attempts, last_error.empty() ? raw : last_error)});
}
throw std::runtime_error("LlamaGenerator: malformed user response");
}