def write_conversations_format(self, outfile, world):
Conversations.save_conversations(
self._logs,
outfile,
world.opt,
self_chat=world.opt.get('selfchat_task', False),
)
def write_conversations_format(self, outfile, world):
logging.info(f'Saving log to {outfile} in Conversations format')
Conversations.save_conversations(
self._logs,
outfile,
world.opt,
self_chat=world.opt.get('selfchat_task', False),
)
def test_conversations(self):
act_list = [
[
[
{'id': 'Emily', 'text': 'Hello, do you like this test?'},
{'id': 'Stephen', 'text': 'Why yes! I love this test!'},
],
[
{'id': 'Emily', 'text': 'So will you stamp this diff?'},
{'id': 'Stephen', 'text': 'Yes, I will do it right now!'},
],
],
[
[
{
'id': 'A',
'text': 'Somebody once told me the world is gonna roll me',
},
{'id': 'B', 'text': 'I aint the sharpest tool in the shed'},
],
[
{
'id': 'A',
'text': 'She was looking kind of dumb with her finger and her thumb',
},
{'id': 'B', 'text': 'In the shape of an L on her forehead'},
],
],
]
self.opt = {
'A': 'B',
'C': 'D',
'E': 'F',
}
self.convo_datapath = os.path.join(self.datapath, 'convo1')
Conversations.save_conversations(
act_list,
self.convo_datapath,
self.opt,
self_chat=False,
other_info='Blah blah blah',
)
assert os.path.exists(self.convo_datapath + '.jsonl')
assert os.path.exists(self.convo_datapath + '.metadata')
convos = Conversations(self.convo_datapath + '.jsonl')
# test conversations loaded
self.assertEqual(convos.num_conversations, 2)
# test speakers saved
speakers = {'Stephen', 'Emily', 'A', 'B'}
self.assertEqual(set(convos.metadata.speakers), speakers)
# test opt saved
for x in ['A', 'C', 'E']:
self.assertEqual(
self.opt[x], convos.metadata.opt[x],
)
# test kwargs
self.assertEqual({'other_info': 'Blah blah blah'}, convos.metadata.extra_data)
# test reading conversations
with testing_utils.capture_output() as out:
convos.read_conv_idx(0)
str_version = (
'Emily: Hello, do you like this test?\n'
'Stephen: Why yes! I love this test!\n'
'Emily: So will you stamp this diff?\n'
'Stephen: Yes, I will do it right now!\n'
)
self.assertIn(str_version, out.getvalue())
def test_conversations(self):
act_list = [
[
[
{
'id': 'Emily',
'text': 'Hello, do you like this test?'
},
{
'id': 'Stephen',
'text': 'Why yes! I love this test!'
},
],
[
{
'id': 'Emily',
'text': 'So will you stamp this diff?'
},
{
'id': 'Stephen',
'text': 'Yes, I will do it right now!'
},
],
],
[
[
{
'id': 'A',
'text':
'Somebody once told me the world is gonna roll me',
},
{
'id': 'B',
'text': 'I aint the sharpest tool in the shed'
},
],
[
{
'id':
'A',
'text':
'She was looking kind of dumb with her finger and her thumb',
},
{
'id': 'B',
'text': 'In the shape of an L on her forehead'
},
],
],
]
self.opt = {'A': 'B', 'C': 'D', 'E': 'F'}
self.convo_datapath = os.path.join(self.datapath, 'convo1')
Conversations.save_conversations(
act_list,
self.convo_datapath,
self.opt,
self_chat=False,
other_info='Blah blah blah',
)
assert os.path.exists(self.convo_datapath + '.jsonl')
assert os.path.exists(self.convo_datapath + '.metadata')
convos = Conversations(self.convo_datapath + '.jsonl')
# test conversations loaded
self.assertEqual(len(convos), 2)
# test speakers saved
speakers = {'Stephen', 'Emily', 'A', 'B'}
self.assertEqual(set(convos.metadata.speakers), speakers)
# test opt saved
for x in ['A', 'C', 'E']:
self.assertEqual(self.opt[x], convos.metadata.opt[x])
# test kwargs
self.assertEqual({'other_info': 'Blah blah blah'},
convos.metadata.extra_data)
# test reading conversations
with self.assertLogs(logger=logging.logger, level='DEBUG') as cm:
convos.read_conv_idx(0)
str_version = ('Emily: Hello, do you like this test?\n'
'Stephen: Why yes! I love this test!\n'
'Emily: So will you stamp this diff?\n'
'Stephen: Yes, I will do it right now!\n')
self.assertIn(str_version, "\n".join(cm.output))
# test getting a specific turn
first = convos[0] # Conversation
self.assertEqual(first[0].id, 'Emily')
self.assertEqual(first[3].text, 'Yes, I will do it right now!')
def dump_data(opt):
"""
Dump task data to ACUTE-Eval.
"""
# create repeat label agent and assign it to the specified task
agent = RepeatLabelAgent(opt)
world = create_task(opt, agent)
task = opt.get('task')
speaker_0_id = opt.get('speaker_0_id') or f'{task}_as_human'
speaker_1_id = opt.get('speaker_1_id') or f'{task}_as_model'
if opt['outfile'] is None:
outfile = tempfile.mkstemp(prefix='{}_{}_'.format(
opt['task'], opt['datatype']),
suffix='.txt')[1]
else:
outfile = opt['outfile']
num_episodes = (world.num_episodes() if opt['num_episodes'] == -1 else min(
opt['num_episodes'], world.num_episodes()))
log_timer = TimeLogger()
print(f'[ starting to convert, saving output to {outfile} ]')
dialogues = []
for _ in range(num_episodes):
episode = []
episode_done = False
while not episode_done:
world.parley()
acts = world.get_acts()
text = acts[0].get('text')
split_text = text.split('\n')
label = random.choice(acts[0].get('labels',
acts[0].pop('eval_labels',
None)))
if not episode and opt.get('prepended_context'):
# first turn
context = split_text[:-1]
text = split_text[-1]
context_turn = [{
'text': context,
'episode_done': False,
'id': 'context'
} for _ in range(2)]
episode.append(context_turn)
turn = [
{
'text': text,
'episode_done': False,
'id': speaker_0_id
},
{
'text': label,
'episode_done': False,
'id': speaker_1_id
},
]
episode.append(turn)
if acts[0].get('episode_done', False):
episode[-1][-1]['episode_done'] = True
episode_done = True
dialogues.append(episode)
if log_timer.time() > opt['log_every_n_secs']:
text, _log = log_timer.log(world.total_parleys,
world.num_examples())
print(text)
if world.epoch_done():
break
Conversations.save_conversations(dialogues, outfile, opt)