Skip to content

Commit

Permalink
ci fix
Browse files Browse the repository at this point in the history
  • Loading branch information
@sbalandi
sbalandi committed Jan 14, 2025
1 parent 11dec94 commit f5d74b4
Show file tree
Hide file tree
Showing 2 changed files with 11 additions and 11 deletions.
20 changes: 10 additions & 10 deletions .github/workflows/causal_lm_cpp.yml
View file
Original file line number Diff line number Diff line change
Expand Up @@ -122,7 +122,7 @@ jobs:
prompt = 'Why is the Sun yellow?'
if tokenizer.chat_template:
prompt = tokenizer.apply_chat_template([{'role': 'user', 'content': prompt}], tokenize=False, add_generation_prompt=True)
tokenized = tokenizer(prompt, return_tensors='pt')
tokenized = tokenizer(prompt, return_tensors='pt', add_special_tokens=False)
for beam in transformers.LlamaForCausalLM.from_pretrained('TinyLlama/TinyLlama-1.1B-Chat-v1.0').generate(**tokenized, num_beam_groups=3, num_beams=15, num_return_sequences=15, diversity_penalty=1.0, max_new_tokens=20, early_stopping=False, length_penalty=1.0, no_repeat_ngram_size=9**9, do_sample=False):
ref = ': ' + tokenizer.decode(beam[tokenized['input_ids'].numel():], skip_special_tokens=True)
idx = predictions.find(ref)
Expand All @@ -141,7 +141,7 @@ jobs:
prompt = '69'
if tokenizer.chat_template:
prompt = tokenizer.apply_chat_template([{'role': 'user', 'content': prompt}], tokenize=False, add_generation_prompt=True)
tokenized = tokenizer(prompt, return_tensors='pt')
tokenized = tokenizer(prompt, return_tensors='pt', add_special_tokens=False)
for beam in transformers.LlamaForCausalLM.from_pretrained('TinyLlama/TinyLlama-1.1B-Chat-v1.0').generate(**tokenized, num_beam_groups=3, num_beams=15, num_return_sequences=15, diversity_penalty=1.0, max_new_tokens=20, early_stopping=False, length_penalty=1.0, no_repeat_ngram_size=9**9, do_sample=False):
ref = ': ' + tokenizer.decode(beam[tokenized['input_ids'].numel():], skip_special_tokens=True)
idx = predictions.find(ref)
Expand All @@ -160,7 +160,7 @@ jobs:
prompt = 'Hi'
if tokenizer.chat_template:
prompt = tokenizer.apply_chat_template([{'role': 'user', 'content': prompt}], tokenize=False, add_generation_prompt=True)
tokenized = tokenizer(prompt, return_tensors='pt')
tokenized = tokenizer(prompt, return_tensors='pt', add_special_tokens=False)
for beam in transformers.LlamaForCausalLM.from_pretrained('TinyLlama/TinyLlama-1.1B-Chat-v1.0').generate(**tokenized, num_beam_groups=3, num_beams=15, num_return_sequences=15, diversity_penalty=1.0, max_new_tokens=20, early_stopping=False, length_penalty=1.0, no_repeat_ngram_size=9**9, do_sample=False):
ref = ': ' + tokenizer.decode(beam[tokenized['input_ids'].numel():], skip_special_tokens=True)
idx = predictions.find(ref)
Expand All @@ -179,7 +179,7 @@ jobs:
prompt = 'return 0'
if tokenizer.chat_template:
prompt = tokenizer.apply_chat_template([{'role': 'user', 'content': prompt}], tokenize=False, add_generation_prompt=True)
tokenized = tokenizer(prompt, return_tensors='pt')
tokenized = tokenizer(prompt, return_tensors='pt', add_special_tokens=False)
for beam in transformers.LlamaForCausalLM.from_pretrained('TinyLlama/TinyLlama-1.1B-Chat-v1.0').generate(**tokenized, num_beam_groups=3, num_beams=15, num_return_sequences=15, diversity_penalty=1.0, max_new_tokens=20, early_stopping=False, length_penalty=1.0, no_repeat_ngram_size=9**9, do_sample=False):
ref = ': ' + tokenizer.decode(beam[tokenized['input_ids'].numel():], skip_special_tokens=True)
idx = predictions.find(ref)
Expand All @@ -198,7 +198,7 @@ jobs:
prompt = '你好！ 你好嗎？'
if tokenizer.chat_template:
prompt = tokenizer.apply_chat_template([{'role': 'user', 'content': prompt}], tokenize=False, add_generation_prompt=True)
tokenized = tokenizer(prompt, return_tensors='pt')
tokenized = tokenizer(prompt, return_tensors='pt', add_special_tokens=False)
for beam in transformers.LlamaForCausalLM.from_pretrained('TinyLlama/TinyLlama-1.1B-Chat-v1.0').generate(**tokenized, num_beam_groups=3, num_beams=15, num_return_sequences=15, diversity_penalty=1.0, max_new_tokens=20, early_stopping=False, length_penalty=1.0, no_repeat_ngram_size=9**9, do_sample=False):
ref = ': ' + tokenizer.decode(beam[tokenized['input_ids'].numel():], skip_special_tokens=True)
idx = predictions.find(ref.replace('�', ''))
Expand All @@ -222,7 +222,7 @@ jobs:
for prompt in prompts:
if tokenizer.chat_template:
prompt = tokenizer.apply_chat_template([{'role': 'user', 'content': prompt}], tokenize=False, add_generation_prompt=True)
tokenized = tokenizer(prompt, return_tensors='pt')
tokenized = tokenizer(prompt, return_tensors='pt', add_special_tokens=False)
for beam in transformers.LlamaForCausalLM.from_pretrained('TinyLlama/TinyLlama-1.1B-Chat-v1.0').generate(**tokenized, num_beam_groups=3, num_beams=15, num_return_sequences=15, diversity_penalty=1.0, max_new_tokens=20, early_stopping=False, length_penalty=1.0, no_repeat_ngram_size=9**9, do_sample=False):
ref = ': ' + tokenizer.decode(beam[tokenized['input_ids'].numel():], skip_special_tokens=True)
idx = predictions.find(ref.replace('�', ''))
Expand Down Expand Up @@ -272,9 +272,9 @@ jobs:
echo predictions = open('cpp.txt', 'r').read() >> ref.py
echo tokenizer = transformers.AutoTokenizer.from_pretrained('TinyLlama/TinyLlama-1.1B-Chat-v1.0', trust_remote_code=True) >> ref.py
echo prompt = '69'
echo if tokenizer.chat_template:
echo if tokenizer.chat_template:
echo prompt = tokenizer.apply_chat_template([{'role': 'user', 'content': prompt}], tokenize=False, add_generation_prompt=True)
echo tokenized = tokenizer(prompt, return_tensors='pt') >> ref.py
echo tokenized = tokenizer(prompt, return_tensors='pt', add_special_tokens=False) >> ref.py
echo for beam in transformers.AutoModelForCausalLM.from_pretrained('TinyLlama/TinyLlama-1.1B-Chat-v1.0', trust_remote_code=True).generate(**tokenized, max_new_tokens=100, do_sample=False): >> ref.py
echo ref = tokenizer.decode(beam[tokenized['input_ids'].numel():], skip_special_tokens=True) >> ref.py
echo idx = predictions.find(ref) >> ref.py
Expand Down Expand Up @@ -582,7 +582,7 @@ jobs:
prompt = 'Alan Turing was a'
if tokenizer.chat_template:
prompt = tokenizer.apply_chat_template([{'role': 'user', 'content': prompt}], tokenize=False, add_generation_prompt=True)
tokenized = tokenizer(prompt, return_tensors='pt')
tokenized = tokenizer(prompt, return_tensors='pt', add_special_tokens=False)
for output in transformers.AutoModelForCausalLM.from_pretrained('microsoft/phi-1_5').generate(**tokenized, max_length=100, do_sample=False):
ref = tokenizer.decode(output[tokenized['input_ids'].numel():], skip_special_tokens=True)
idx = predictions.find(ref)
Expand Down Expand Up @@ -640,7 +640,7 @@ jobs:
prompt = 'Alan Turing was a'
if tokenizer.chat_template:
prompt = tokenizer.apply_chat_template([{'role': 'user', 'content': prompt}], tokenize=False, add_generation_prompt=True)
tokenized = tokenizer(prompt, return_tensors='pt')
tokenized = tokenizer(prompt, return_tensors='pt', add_special_tokens=False)
for output in transformers.AutoModelForCausalLM.from_pretrained('ikala/redpajama-3b-chat').generate(**tokenized, max_length=100, do_sample=False):
ref = tokenizer.decode(output[tokenized['input_ids'].numel():], skip_special_tokens=True)
idx = predictions.find(ref)
Expand Down
2 changes: 1 addition & 1 deletion tests/python_tests/common.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -274,7 +274,7 @@ def run_hugging_face(
else:
processed_prompts = prompts
# process all prompts as a single batch as we have a single generation config for all prompts
inputs = hf_tokenizer(processed_prompts, return_tensors='pt', padding=True, truncation=True, add_special_tokens=True, padding_side='left')
inputs = hf_tokenizer(processed_prompts, return_tensors='pt', padding=True, truncation=True, add_special_tokens=False, padding_side='left')
input_ids, attention_mask = inputs['input_ids'], inputs['attention_mask']
hf_generation_config = convert_to_hf(opt_model.generation_config, generation_configs)
hf_encoded_outputs = opt_model.generate(input_ids, attention_mask=attention_mask, generation_config=hf_generation_config, tokenizer=hf_tokenizer)
Expand Down

0 comments on commit f5d74b4

Please sign in to comment.