Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Enable the API cost tracking in Synthesizer #1406

Merged
merged 3 commits into from
Mar 3, 2025
Merged

Enable the API cost tracking in Synthesizer #1406

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
874bab0
enable cost tracking for synthesizer
chuqingG Mar 1, 2025
f6bf569
Merge branch 'main' of github.com:chuqingG/deepeval
chuqingG Mar 1, 2025
499c64c
reformat
chuqingG Mar 1, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 7 additions & 2 deletions deepeval/synthesizer/chunking/context_generator.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -67,6 +67,9 @@ def __init__(
Dict[str, Collection]
] = None

# cost tracking
self.total_cost = 0.0

#########################################################
### Generate Contexts ###################################
#########################################################
Expand Down Expand Up @@ -499,7 +502,8 @@ async def a_evaluate_chunk_and_update(chunk):
def evaluate_chunk(self, chunk) -> float:
prompt = FilterTemplate.evaluate_context(chunk)
if self.using_native_model:
res, _ = self.model.generate(prompt, schema=ContextScore)
res, cost = self.model.generate(prompt, schema=ContextScore)
self.total_cost += cost
return (res.clarity + res.depth + res.structure + res.relevance) / 4
else:
try:
Expand All @@ -523,7 +527,8 @@ def evaluate_chunk(self, chunk) -> float:
async def a_evaluate_chunk(self, chunk) -> float:
prompt = FilterTemplate.evaluate_context(chunk)
if self.using_native_model:
res, _ = await self.model.a_generate(prompt, schema=ContextScore)
res, cost = await self.model.a_generate(prompt, schema=ContextScore)
self.total_cost += cost
return (res.clarity + res.depth + res.structure + res.relevance) / 4
else:

Expand Down
31 changes: 27 additions & 4 deletions deepeval/synthesizer/synthesizer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -81,6 +81,7 @@ def __init__(
filtration_config: Optional[FiltrationConfig] = None,
evolution_config: Optional[EvolutionConfig] = None,
styling_config: Optional[StylingConfig] = None,
cost_tracking: bool = False,
):
self.model, self.using_native_model = initialize_model(model)
self.async_mode = async_mode
Expand All @@ -100,6 +101,8 @@ def __init__(
self.styling_config = (
styling_config if styling_config is not None else StylingConfig()
)
self.cost_tracking = cost_tracking
self.synthesis_cost = 0 if self.using_native_model else None

#############################################################
# Generate Goldens from Docs
Expand Down Expand Up @@ -127,6 +130,7 @@ def generate_goldens_from_docs(
include_expected_output=include_expected_output,
max_goldens_per_context=max_goldens_per_context,
context_construction_config=context_construction_config,
_reset_cost=False,
)
)
else:
Expand All @@ -149,6 +153,8 @@ def generate_goldens_from_docs(
max_context_size=context_construction_config.max_context_length,
)
)
if self.synthesis_cost:
self.synthesis_cost += self.context_generator.total_cost
print(
f"Utilizing {len(set(chain.from_iterable(contexts)))} out of {self.context_generator.total_chunks} chunks."
)
Expand All @@ -170,8 +176,10 @@ def generate_goldens_from_docs(
_context_scores=context_scores,
_progress_bar=progress_bar,
_send_data=False,
_reset_cost=False,
)

if self.cost_tracking and self.using_native_model:
print(f"💰 API cost: {self.synthesis_cost:.6f}")
# Wrap-up Synthesis
if _send_data == True:
pass
Expand All @@ -183,12 +191,14 @@ async def a_generate_goldens_from_docs(
include_expected_output: bool = True,
max_goldens_per_context: int = 2,
context_construction_config: Optional[ContextConstructionConfig] = None,
_reset_cost=True,
):
if context_construction_config is None:
context_construction_config = ContextConstructionConfig(
critic_model=self.model
)
self.synthesis_cost = 0 if self.using_native_model else None
if _reset_cost:
self.synthesis_cost = 0 if self.using_native_model else None

# Generate contexts from provided docs
if self.context_generator is None:
Expand All @@ -210,6 +220,8 @@ async def a_generate_goldens_from_docs(
max_context_size=context_construction_config.max_context_length,
)
)
if self.synthesis_cost:
self.synthesis_cost += self.context_generator.total_cost
print(
f"Utilizing {len(set(chain.from_iterable(contexts)))} out of {self.context_generator.total_chunks} chunks."
)
Expand All @@ -230,8 +242,11 @@ async def a_generate_goldens_from_docs(
source_files=source_files,
_context_scores=context_scores,
_progress_bar=progress_bar,
_reset_cost=False,
)
self.synthetic_goldens.extend(goldens)
if _reset_cost and self.cost_tracking and self.using_native_model:
print(f"💰 API cost: {self.synthesis_cost:.6f}")
return goldens

#############################################################
Expand All @@ -247,8 +262,10 @@ def generate_goldens_from_contexts(
_context_scores: Optional[List[float]] = None,
_progress_bar: Optional[tqdm.std.tqdm] = None,
_send_data: bool = True,
_reset_cost: bool = True,
) -> List[Golden]:
self.synthesis_cost = 0 if self.using_native_model else None
if _reset_cost:
self.synthesis_cost = 0 if self.using_native_model else None
# Intialize Goldens as an empty list
goldens: List[Golden] = []

Expand Down Expand Up @@ -358,6 +375,8 @@ def generate_goldens_from_contexts(
self.synthetic_goldens.extend(goldens)
if _send_data == True:
pass
if _reset_cost and self.cost_tracking and self.using_native_model:
print(f"💰 API cost: {self.synthesis_cost:.6f}")
return goldens

async def a_generate_goldens_from_contexts(
Expand All @@ -368,8 +387,10 @@ async def a_generate_goldens_from_contexts(
source_files: Optional[List[str]] = None,
_context_scores: Optional[List[float]] = None,
_progress_bar: Optional[tqdm.std.tqdm] = None,
_reset_cost: bool = True,
) -> List[Golden]:
self.synthesis_cost = 0 if self.using_native_model else None
if _reset_cost:
self.synthesis_cost = 0 if self.using_native_model else None
semaphore = asyncio.Semaphore(self.max_concurrent)
goldens: List[Golden] = []
with synthesizer_progress_context(
Expand Down Expand Up @@ -399,6 +420,8 @@ async def a_generate_goldens_from_contexts(
]
await asyncio.gather(*tasks)

if _reset_cost and self.cost_tracking and self.using_native_model:
print(f"💰 API cost: {self.synthesis_cost:.6f}")
return goldens

async def _a_generate_from_context(
Expand Down
Loading