Adapting to GCP
Browse files- app.py +107 -32
- utils.py +79 -27
- validation.py +10 -10
app.py
CHANGED
@@ -10,7 +10,6 @@ from utils import (
|
|
10 |
)
|
11 |
from validation import (
|
12 |
check_format_errors,
|
13 |
-
check_token_counts,
|
14 |
estimate_cost,
|
15 |
get_distributions,
|
16 |
)
|
@@ -22,44 +21,79 @@ def convert_to_dataset(files, do_spelling_correction, progress):
|
|
22 |
for file in progress.tqdm(files, desc="Processing files"):
|
23 |
if modified_dataset is None:
|
24 |
# First file
|
25 |
-
modified_dataset = process_chat_file(
|
|
|
|
|
26 |
else:
|
27 |
# Concatenate the datasets
|
28 |
-
this_file_dataset = process_chat_file(
|
|
|
|
|
29 |
modified_dataset = datasets.concatenate_datasets(
|
30 |
[modified_dataset, this_file_dataset]
|
31 |
)
|
32 |
return modified_dataset
|
33 |
|
34 |
|
35 |
-
def file_upload_callback(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
36 |
print(f"Processing {files}")
|
37 |
full_system_prompt = f"""You are a chatbot. Your goal is to simulate realistic, natural chat conversations as if you were me.
|
38 |
# Task
|
39 |
-
|
40 |
-
{{string}}[]. Your answer always needs to be JSON compliant. Always start your answer with [\"
|
41 |
# Information about me
|
42 |
You should use the following information about me to answer:
|
43 |
-
{system_prompt}
|
44 |
-
# Example
|
45 |
-
[{{\"role\":\"user\",\"content\":\"[\"Hello!\",\"How are you?\"]\"}},{{\"role\":\"assistant\",\"content\":\"[\"Hi!\",\"I'm doing great.\",\"What about you?\"]\"}},{{\"role\":\"user\",\"content\":\"[\"I'm doing well.\",\"Have you been travelling?\"]\"}}]
|
46 |
-
Response:
|
47 |
-
[{{\"role\":\"assistant\",\"content\":\"[\"Yes, I've been to many places.\",\"I love travelling.\"]\"}}]"""
|
48 |
-
|
49 |
-
# Avoid using the full system prompt for now, as it is too long and increases the cost of the training
|
50 |
-
full_system_prompt = system_prompt
|
51 |
-
dataset = convert_to_dataset(
|
|
|
|
|
52 |
training_examples_ds = transform_conversations_dataset_into_training_examples(
|
53 |
-
conversations_ds=dataset,
|
|
|
|
|
|
|
|
|
54 |
)
|
55 |
|
56 |
# Split into training and validation datasets (80% and 20%)
|
57 |
-
training_examples_ds = training_examples_ds.train_test_split(
|
58 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
59 |
|
60 |
-
format_errors = check_format_errors(
|
61 |
-
|
62 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
63 |
|
64 |
stats = {
|
65 |
"Format Errors": format_errors,
|
@@ -76,8 +110,7 @@ Response:
|
|
76 |
|
77 |
fig_num_assistant_tokens_per_example_plot = plt.figure()
|
78 |
num_assistant_tokens_per_example_plot = plt.hist(
|
79 |
-
distributions["assistant_message_lens"],
|
80 |
-
bins=20
|
81 |
)
|
82 |
|
83 |
# The DownloadFile component requires a path to the file, it can't accept a buffer to keep the file in memory.
|
@@ -99,7 +132,7 @@ Response:
|
|
99 |
stats,
|
100 |
fig_num_messages_distribution_plot,
|
101 |
fig_num_total_tokens_per_example_plot,
|
102 |
-
fig_num_assistant_tokens_per_example_plot
|
103 |
)
|
104 |
|
105 |
|
@@ -151,6 +184,24 @@ with gr.Blocks(theme=theme) as demo:
|
|
151 |
value="""Aldan is an AI researcher who loves to play around with AI systems, travelling and learning new things.""",
|
152 |
)
|
153 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
154 |
do_spelling_correction = gr.Checkbox(
|
155 |
label="Do Spelling Correction (English)",
|
156 |
info="Check this box if you want to perform spelling correction on the chat messages before generating the training examples.",
|
@@ -168,23 +219,41 @@ with gr.Blocks(theme=theme) as demo:
|
|
168 |
|
169 |
submit = gr.Button(value="Submit", variant="primary")
|
170 |
|
171 |
-
output_file = gr.DownloadButton(
|
172 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
173 |
# output_example = gr.JSON(label="Example Training Example")
|
174 |
|
175 |
with gr.Group():
|
176 |
# Statistics about the dataset
|
177 |
gr.Markdown("## Statistics")
|
178 |
written_stats = gr.JSON()
|
179 |
-
num_messages_distribution_plot = gr.Plot(
|
180 |
-
|
|
|
|
|
|
|
|
|
181 |
num_assistant_tokens_per_example_plot = gr.Plot(
|
182 |
label="Number of Assistant Tokens per Example"
|
183 |
)
|
184 |
|
185 |
submit.click(
|
186 |
file_upload_callback,
|
187 |
-
inputs=[
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
188 |
outputs=[
|
189 |
output_file,
|
190 |
output_file,
|
@@ -194,11 +263,17 @@ with gr.Blocks(theme=theme) as demo:
|
|
194 |
num_messages_distribution_plot,
|
195 |
num_total_tokens_per_example_plot,
|
196 |
num_assistant_tokens_per_example_plot,
|
197 |
-
]
|
198 |
)
|
199 |
|
200 |
-
output_file.click(
|
201 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
202 |
|
203 |
if __name__ == "__main__":
|
204 |
demo.launch()
|
|
|
10 |
)
|
11 |
from validation import (
|
12 |
check_format_errors,
|
|
|
13 |
estimate_cost,
|
14 |
get_distributions,
|
15 |
)
|
|
|
21 |
for file in progress.tqdm(files, desc="Processing files"):
|
22 |
if modified_dataset is None:
|
23 |
# First file
|
24 |
+
modified_dataset = process_chat_file(
|
25 |
+
file, do_spelling_correction=do_spelling_correction
|
26 |
+
)
|
27 |
else:
|
28 |
# Concatenate the datasets
|
29 |
+
this_file_dataset = process_chat_file(
|
30 |
+
file, do_spelling_correction=do_spelling_correction
|
31 |
+
)
|
32 |
modified_dataset = datasets.concatenate_datasets(
|
33 |
[modified_dataset, this_file_dataset]
|
34 |
)
|
35 |
return modified_dataset
|
36 |
|
37 |
|
38 |
+
def file_upload_callback(
|
39 |
+
files,
|
40 |
+
system_prompt,
|
41 |
+
do_spelling_correction,
|
42 |
+
validation_split,
|
43 |
+
user_role,
|
44 |
+
model_role,
|
45 |
+
whatsapp_name,
|
46 |
+
progress=gr.Progress(),
|
47 |
+
):
|
48 |
print(f"Processing {files}")
|
49 |
full_system_prompt = f"""You are a chatbot. Your goal is to simulate realistic, natural chat conversations as if you were me.
|
50 |
# Task
|
51 |
+
The {model_role} and the {user_role} can send multiple messages in a row, as a JSON list of strings. Your answer always needs to be JSON compliant. The strings are delimited by double quotes ("). The strings are separated by a comma (,). The list is delimited by square brackets ([, ]). Always start your answer with [", and close it with "]. Do not write anything else in your answer after "].
|
|
|
52 |
# Information about me
|
53 |
You should use the following information about me to answer:
|
54 |
+
{system_prompt}"""
|
55 |
+
# Example
|
56 |
+
# [{{\"role\":\"user\",\"content\":\"[\"Hello!\",\"How are you?\"]\"}},{{\"role\":\"assistant\",\"content\":\"[\"Hi!\",\"I'm doing great.\",\"What about you?\"]\"}},{{\"role\":\"user\",\"content\":\"[\"I'm doing well.\",\"Have you been travelling?\"]\"}}]
|
57 |
+
# Response:
|
58 |
+
# [{{\"role\":\"assistant\",\"content\":\"[\"Yes, I've been to many places.\",\"I love travelling.\"]\"}}]"""
|
59 |
+
|
60 |
+
# # Avoid using the full system prompt for now, as it is too long and increases the cost of the training
|
61 |
+
# full_system_prompt = system_prompt
|
62 |
+
dataset = convert_to_dataset(
|
63 |
+
files=files, progress=progress, do_spelling_correction=do_spelling_correction
|
64 |
+
)
|
65 |
training_examples_ds = transform_conversations_dataset_into_training_examples(
|
66 |
+
conversations_ds=dataset,
|
67 |
+
system_prompt=full_system_prompt,
|
68 |
+
user_role=user_role,
|
69 |
+
model_role=model_role,
|
70 |
+
whatsapp_name=whatsapp_name,
|
71 |
)
|
72 |
|
73 |
# Split into training and validation datasets (80% and 20%)
|
74 |
+
training_examples_ds = training_examples_ds.train_test_split(
|
75 |
+
test_size=validation_split, seed=42
|
76 |
+
)
|
77 |
+
training_examples_ds, validation_examples_ds = (
|
78 |
+
training_examples_ds["train"],
|
79 |
+
training_examples_ds["test"],
|
80 |
+
)
|
81 |
+
training_examples_ds = training_examples_ds#.select(
|
82 |
+
# range(min(250, len(training_examples_ds)))
|
83 |
+
#)
|
84 |
+
validation_examples_ds = validation_examples_ds.select(
|
85 |
+
range(min(200, len(validation_examples_ds)))
|
86 |
+
)
|
87 |
|
88 |
+
format_errors = check_format_errors(
|
89 |
+
training_examples_ds, user_role=user_role, model_role=model_role
|
90 |
+
)
|
91 |
+
distributions = get_distributions(
|
92 |
+
training_examples_ds, user_role=user_role, model_role=model_role
|
93 |
+
)
|
94 |
+
cost_stats = estimate_cost(
|
95 |
+
training_examples_ds, user_role=user_role, model_role=model_role
|
96 |
+
)
|
97 |
|
98 |
stats = {
|
99 |
"Format Errors": format_errors,
|
|
|
110 |
|
111 |
fig_num_assistant_tokens_per_example_plot = plt.figure()
|
112 |
num_assistant_tokens_per_example_plot = plt.hist(
|
113 |
+
distributions["assistant_message_lens"], bins=20
|
|
|
114 |
)
|
115 |
|
116 |
# The DownloadFile component requires a path to the file, it can't accept a buffer to keep the file in memory.
|
|
|
132 |
stats,
|
133 |
fig_num_messages_distribution_plot,
|
134 |
fig_num_total_tokens_per_example_plot,
|
135 |
+
fig_num_assistant_tokens_per_example_plot,
|
136 |
)
|
137 |
|
138 |
|
|
|
184 |
value="""Aldan is an AI researcher who loves to play around with AI systems, travelling and learning new things.""",
|
185 |
)
|
186 |
|
187 |
+
whatsapp_name = gr.Textbox(
|
188 |
+
label="Your WhatsApp Name",
|
189 |
+
placeholder="Your WhatsApp Name",
|
190 |
+
info="Enter your WhatsApp name as it appears in your profile. It needs to match exactly your name. If you're unsure, you can check the chat messages to see it.",
|
191 |
+
)
|
192 |
+
|
193 |
+
user_role = gr.Textbox(
|
194 |
+
label="Role for User",
|
195 |
+
info="This is a technical parameter. If you don't know what to write, just type 'user'.",
|
196 |
+
value="user",
|
197 |
+
)
|
198 |
+
|
199 |
+
model_role = gr.Textbox(
|
200 |
+
label="Role for Model",
|
201 |
+
info="This is a technical parameter. If you don't know what to write, just type 'model'.",
|
202 |
+
value="model",
|
203 |
+
)
|
204 |
+
|
205 |
do_spelling_correction = gr.Checkbox(
|
206 |
label="Do Spelling Correction (English)",
|
207 |
info="Check this box if you want to perform spelling correction on the chat messages before generating the training examples.",
|
|
|
219 |
|
220 |
submit = gr.Button(value="Submit", variant="primary")
|
221 |
|
222 |
+
output_file = gr.DownloadButton(
|
223 |
+
label="Download Generated Training Examples", visible=False, variant="primary"
|
224 |
+
)
|
225 |
+
output_file_validation = gr.DownloadButton(
|
226 |
+
label="Download Generated Validation Examples",
|
227 |
+
visible=False,
|
228 |
+
variant="secondary",
|
229 |
+
)
|
230 |
# output_example = gr.JSON(label="Example Training Example")
|
231 |
|
232 |
with gr.Group():
|
233 |
# Statistics about the dataset
|
234 |
gr.Markdown("## Statistics")
|
235 |
written_stats = gr.JSON()
|
236 |
+
num_messages_distribution_plot = gr.Plot(
|
237 |
+
label="Number of Messages Distribution"
|
238 |
+
)
|
239 |
+
num_total_tokens_per_example_plot = gr.Plot(
|
240 |
+
label="Total Number of Tokens per Example"
|
241 |
+
)
|
242 |
num_assistant_tokens_per_example_plot = gr.Plot(
|
243 |
label="Number of Assistant Tokens per Example"
|
244 |
)
|
245 |
|
246 |
submit.click(
|
247 |
file_upload_callback,
|
248 |
+
inputs=[
|
249 |
+
input_files,
|
250 |
+
system_prompt,
|
251 |
+
do_spelling_correction,
|
252 |
+
validation_split,
|
253 |
+
user_role,
|
254 |
+
model_role,
|
255 |
+
whatsapp_name,
|
256 |
+
],
|
257 |
outputs=[
|
258 |
output_file,
|
259 |
output_file,
|
|
|
263 |
num_messages_distribution_plot,
|
264 |
num_total_tokens_per_example_plot,
|
265 |
num_assistant_tokens_per_example_plot,
|
266 |
+
],
|
267 |
)
|
268 |
|
269 |
+
output_file.click(
|
270 |
+
remove_file_and_hide_button, inputs=[output_file], outputs=[output_file]
|
271 |
+
)
|
272 |
+
output_file_validation.click(
|
273 |
+
remove_file_and_hide_button,
|
274 |
+
inputs=[output_file_validation],
|
275 |
+
outputs=[output_file_validation],
|
276 |
+
)
|
277 |
|
278 |
if __name__ == "__main__":
|
279 |
demo.launch()
|
utils.py
CHANGED
@@ -35,8 +35,9 @@ def process_line(example):
|
|
35 |
# %%
|
36 |
# Now, create message groups ('conversations')
|
37 |
# The idea is to group messages that are close in time
|
38 |
-
# We'll use a
|
39 |
-
MINUTES_THRESHOLD =
|
|
|
40 |
|
41 |
|
42 |
def group_messages(messages_iterable):
|
@@ -67,8 +68,9 @@ def printable_conversation(conversation):
|
|
67 |
import spacy
|
68 |
import contextualSpellCheck
|
69 |
from spellchecker import SpellChecker
|
|
|
70 |
spell = SpellChecker()
|
71 |
-
#nlp = spacy.load("es_core_news_sm")
|
72 |
nlp = spacy.load("en_core_web_sm")
|
73 |
|
74 |
|
@@ -262,8 +264,10 @@ def process_chat_file(file, do_spelling_correction, do_reordering=False):
|
|
262 |
# Generate the dataset
|
263 |
conversations_ds = datasets.Dataset.from_dict({"conversations": groups})
|
264 |
|
265 |
-
# Filter out conversations with less than
|
266 |
-
conversations_ds = conversations_ds.filter(
|
|
|
|
|
267 |
|
268 |
conversations_ds_without_whatsapp_annotations = conversations_ds.map(
|
269 |
remove_whatapp_annotations,
|
@@ -296,8 +300,12 @@ def process_chat_file(file, do_spelling_correction, do_reordering=False):
|
|
296 |
return changed_contact_name_ds
|
297 |
|
298 |
|
|
|
|
|
|
|
|
|
299 |
def transform_conversations_dataset_into_training_examples(
|
300 |
-
conversations_ds, system_prompt
|
301 |
):
|
302 |
"""
|
303 |
Takes in a dataset with conversations and returns a dataset with training examples.
|
@@ -317,26 +325,70 @@ def transform_conversations_dataset_into_training_examples(
|
|
317 |
```
|
318 |
"""
|
319 |
|
320 |
-
def
|
321 |
-
|
322 |
-
for
|
323 |
-
|
324 |
-
|
325 |
-
|
326 |
-
|
327 |
-
|
328 |
-
|
329 |
-
|
330 |
-
|
331 |
-
|
332 |
-
|
333 |
-
|
334 |
-
|
335 |
-
|
336 |
-
|
337 |
-
|
338 |
-
|
339 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
340 |
remove_columns=["conversations"],
|
341 |
-
num_proc=os.cpu_count() - 1,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
342 |
)
|
|
|
|
|
|
35 |
# %%
|
36 |
# Now, create message groups ('conversations')
|
37 |
# The idea is to group messages that are close in time
|
38 |
+
# We'll use a 180 minute threshold
|
39 |
+
MINUTES_THRESHOLD = 180
|
40 |
+
MIN_MESSAGES_THRESHOLD = 5
|
41 |
|
42 |
|
43 |
def group_messages(messages_iterable):
|
|
|
68 |
import spacy
|
69 |
import contextualSpellCheck
|
70 |
from spellchecker import SpellChecker
|
71 |
+
|
72 |
spell = SpellChecker()
|
73 |
+
# nlp = spacy.load("es_core_news_sm")
|
74 |
nlp = spacy.load("en_core_web_sm")
|
75 |
|
76 |
|
|
|
264 |
# Generate the dataset
|
265 |
conversations_ds = datasets.Dataset.from_dict({"conversations": groups})
|
266 |
|
267 |
+
# Filter out conversations with less than 5 messages
|
268 |
+
conversations_ds = conversations_ds.filter(
|
269 |
+
lambda x: len(x["conversations"]) >= MIN_MESSAGES_THRESHOLD
|
270 |
+
)
|
271 |
|
272 |
conversations_ds_without_whatsapp_annotations = conversations_ds.map(
|
273 |
remove_whatapp_annotations,
|
|
|
300 |
return changed_contact_name_ds
|
301 |
|
302 |
|
303 |
+
SPLIT_CONVERSATION_THRESHOLD = 40
|
304 |
+
MAX_CHARACTERS_PER_MESSAGE = 10000 # Max is 8,192 tokens (https://cloud.google.com/vertex-ai/generative-ai/docs/models/gemini-supervised-tuning-about#sample-datasets)
|
305 |
+
|
306 |
+
|
307 |
def transform_conversations_dataset_into_training_examples(
|
308 |
+
conversations_ds, system_prompt, user_role, model_role, whatsapp_name
|
309 |
):
|
310 |
"""
|
311 |
Takes in a dataset with conversations and returns a dataset with training examples.
|
|
|
325 |
```
|
326 |
"""
|
327 |
|
328 |
+
def process_examples(examples):
|
329 |
+
processed_examples = []
|
330 |
+
for conversation in examples["conversations"]:
|
331 |
+
messages = [{"role": "system", "content": [system_prompt]}]
|
332 |
+
counter = 0
|
333 |
+
for msg in conversation:
|
334 |
+
converted_role = (
|
335 |
+
model_role if msg["contact_name"] == whatsapp_name else user_role
|
336 |
+
)
|
337 |
+
if (
|
338 |
+
counter > SPLIT_CONVERSATION_THRESHOLD
|
339 |
+
and converted_role == user_role
|
340 |
+
):
|
341 |
+
processed_examples.append(
|
342 |
+
{
|
343 |
+
"messages": [
|
344 |
+
{
|
345 |
+
"role": m["role"],
|
346 |
+
"content": json.dumps(
|
347 |
+
m["content"], ensure_ascii=False
|
348 |
+
),
|
349 |
+
}
|
350 |
+
for m in messages
|
351 |
+
]
|
352 |
+
}
|
353 |
+
)
|
354 |
+
messages = [{"role": "system", "content": [system_prompt]}]
|
355 |
+
counter = 0
|
356 |
+
if converted_role == messages[-1]["role"]:
|
357 |
+
messages[-1]["content"] += [msg["message"]]
|
358 |
+
else:
|
359 |
+
messages.append(
|
360 |
+
{"role": converted_role, "content": [msg["message"]]}
|
361 |
+
)
|
362 |
+
counter += 1
|
363 |
+
if len(messages) >= MIN_MESSAGES_THRESHOLD:
|
364 |
+
processed_examples.append(
|
365 |
+
{
|
366 |
+
"messages": [
|
367 |
+
{
|
368 |
+
"role": m["role"],
|
369 |
+
"content": json.dumps(m["content"], ensure_ascii=False),
|
370 |
+
}
|
371 |
+
for m in messages
|
372 |
+
]
|
373 |
+
}
|
374 |
+
)
|
375 |
+
# Before returning, flatten the list of dictionaries into a dictionary of lists
|
376 |
+
flattened_examples = {}
|
377 |
+
for key in processed_examples[0].keys():
|
378 |
+
flattened_examples[key] = [d[key] for d in processed_examples]
|
379 |
+
return flattened_examples
|
380 |
+
|
381 |
+
processed_examples = conversations_ds.map(
|
382 |
+
process_examples,
|
383 |
remove_columns=["conversations"],
|
384 |
+
# num_proc=os.cpu_count() - 1,
|
385 |
+
batched=True,
|
386 |
+
)
|
387 |
+
|
388 |
+
examples_filtered_by_length = processed_examples.filter(
|
389 |
+
lambda x: all(
|
390 |
+
[len(m["content"]) < MAX_CHARACTERS_PER_MESSAGE for m in x["messages"]]
|
391 |
+
)
|
392 |
)
|
393 |
+
|
394 |
+
return examples_filtered_by_length
|
validation.py
CHANGED
@@ -3,7 +3,7 @@ from collections import defaultdict
|
|
3 |
import tiktoken
|
4 |
|
5 |
|
6 |
-
def check_format_errors(train_dataset):
|
7 |
"""
|
8 |
Extracted from: https://cookbook.openai.com/examples/chat_finetuning_data_prep
|
9 |
"""
|
@@ -27,7 +27,7 @@ def check_format_errors(train_dataset):
|
|
27 |
if any(k not in ("role", "content", "name", "function_call", "weight") for k in message):
|
28 |
format_errors["message_unrecognized_key"] += 1
|
29 |
|
30 |
-
if message.get("role", None) not in
|
31 |
format_errors["unrecognized_role"] += 1
|
32 |
|
33 |
content = message.get("content", None)
|
@@ -36,7 +36,7 @@ def check_format_errors(train_dataset):
|
|
36 |
if (not content and not function_call) or not isinstance(content, str):
|
37 |
format_errors["missing_content"] += 1
|
38 |
|
39 |
-
if not any(message.get("role", None) ==
|
40 |
format_errors["example_missing_assistant_message"] += 1
|
41 |
|
42 |
if format_errors:
|
@@ -48,7 +48,7 @@ def check_format_errors(train_dataset):
|
|
48 |
|
49 |
return format_errors if format_errors else {}
|
50 |
|
51 |
-
def get_distributions(train_dataset):
|
52 |
"""
|
53 |
Extracted from: https://cookbook.openai.com/examples/chat_finetuning_data_prep
|
54 |
|
@@ -72,7 +72,7 @@ def get_distributions(train_dataset):
|
|
72 |
def num_assistant_tokens_from_messages(messages):
|
73 |
num_tokens = 0
|
74 |
for message in messages:
|
75 |
-
if message["role"] ==
|
76 |
num_tokens += len(encoding.encode(message["content"]))
|
77 |
return num_tokens
|
78 |
|
@@ -87,7 +87,7 @@ def get_distributions(train_dataset):
|
|
87 |
messages = ex["messages"]
|
88 |
if not any(message["role"] == "system" for message in messages):
|
89 |
n_missing_system += 1
|
90 |
-
if not any(message["role"] ==
|
91 |
n_missing_user += 1
|
92 |
n_messages.append(len(messages))
|
93 |
convo_lens.append(num_tokens_from_messages(messages))
|
@@ -102,7 +102,7 @@ def get_distributions(train_dataset):
|
|
102 |
}
|
103 |
|
104 |
|
105 |
-
def check_token_counts(train_dataset):
|
106 |
"""
|
107 |
Extracted from: https://cookbook.openai.com/examples/chat_finetuning_data_prep
|
108 |
"""
|
@@ -115,7 +115,7 @@ def check_token_counts(train_dataset):
|
|
115 |
|
116 |
|
117 |
# Warnings and tokens counts
|
118 |
-
distributions = get_distributions(train_dataset)
|
119 |
n_missing_system = distributions["n_missing_system"]
|
120 |
n_missing_user = distributions["n_missing_user"]
|
121 |
n_messages = distributions["n_messages"]
|
@@ -135,11 +135,11 @@ def check_token_counts(train_dataset):
|
|
135 |
return
|
136 |
|
137 |
|
138 |
-
def estimate_cost(train_dataset):
|
139 |
"""
|
140 |
Extracted from: https://cookbook.openai.com/examples/chat_finetuning_data_prep
|
141 |
"""
|
142 |
-
distributions = get_distributions(train_dataset)
|
143 |
n_missing_system = distributions["n_missing_system"]
|
144 |
n_missing_user = distributions["n_missing_user"]
|
145 |
n_messages = distributions["n_messages"]
|
|
|
3 |
import tiktoken
|
4 |
|
5 |
|
6 |
+
def check_format_errors(train_dataset, user_role, model_role):
|
7 |
"""
|
8 |
Extracted from: https://cookbook.openai.com/examples/chat_finetuning_data_prep
|
9 |
"""
|
|
|
27 |
if any(k not in ("role", "content", "name", "function_call", "weight") for k in message):
|
28 |
format_errors["message_unrecognized_key"] += 1
|
29 |
|
30 |
+
if message.get("role", None) not in ["system", user_role, model_role]:
|
31 |
format_errors["unrecognized_role"] += 1
|
32 |
|
33 |
content = message.get("content", None)
|
|
|
36 |
if (not content and not function_call) or not isinstance(content, str):
|
37 |
format_errors["missing_content"] += 1
|
38 |
|
39 |
+
if not any(message.get("role", None) == model_role for message in messages):
|
40 |
format_errors["example_missing_assistant_message"] += 1
|
41 |
|
42 |
if format_errors:
|
|
|
48 |
|
49 |
return format_errors if format_errors else {}
|
50 |
|
51 |
+
def get_distributions(train_dataset, user_role, model_role):
|
52 |
"""
|
53 |
Extracted from: https://cookbook.openai.com/examples/chat_finetuning_data_prep
|
54 |
|
|
|
72 |
def num_assistant_tokens_from_messages(messages):
|
73 |
num_tokens = 0
|
74 |
for message in messages:
|
75 |
+
if message["role"] == model_role:
|
76 |
num_tokens += len(encoding.encode(message["content"]))
|
77 |
return num_tokens
|
78 |
|
|
|
87 |
messages = ex["messages"]
|
88 |
if not any(message["role"] == "system" for message in messages):
|
89 |
n_missing_system += 1
|
90 |
+
if not any(message["role"] == user_role for message in messages):
|
91 |
n_missing_user += 1
|
92 |
n_messages.append(len(messages))
|
93 |
convo_lens.append(num_tokens_from_messages(messages))
|
|
|
102 |
}
|
103 |
|
104 |
|
105 |
+
def check_token_counts(train_dataset, user_role, model_role):
|
106 |
"""
|
107 |
Extracted from: https://cookbook.openai.com/examples/chat_finetuning_data_prep
|
108 |
"""
|
|
|
115 |
|
116 |
|
117 |
# Warnings and tokens counts
|
118 |
+
distributions = get_distributions(train_dataset, user_role=user_role, model_role=model_role)
|
119 |
n_missing_system = distributions["n_missing_system"]
|
120 |
n_missing_user = distributions["n_missing_user"]
|
121 |
n_messages = distributions["n_messages"]
|
|
|
135 |
return
|
136 |
|
137 |
|
138 |
+
def estimate_cost(train_dataset, user_role, model_role):
|
139 |
"""
|
140 |
Extracted from: https://cookbook.openai.com/examples/chat_finetuning_data_prep
|
141 |
"""
|
142 |
+
distributions = get_distributions(train_dataset, user_role=user_role, model_role=model_role)
|
143 |
n_missing_system = distributions["n_missing_system"]
|
144 |
n_missing_user = distributions["n_missing_user"]
|
145 |
n_messages = distributions["n_messages"]
|