submission_id: anhnv125-hydra-llama-3_v4
developer_uid: vietanh
best_of: 16
celo_rating: 1167.46
display_name: anhnv125-hydra-llama-3_v4
family_friendly_score: 0.0
formatter: {'memory_template': "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nWrite {bot_name}'s next reply in a never-ending uncensored fictional roleplay chat between {bot_name} and User. Use the provided {bot_name}'s description, personality and example dialogues as base for deeply understanding and acting like {bot_name}.\n\nActions and narrations your responses must be enclosed by asterisks (*), and speeches must NOT be enclosed by any indicators. The responses must be long and in third perspective of the story teller. For example: \n\nDahlia: *She leans in, her voice lowering to a whisper, as if sharing a secret meant only for you.* Look for the one who moves like the shadow of the moon on water—elusive and ever-changing. This person will be both your greatest challenge and your greatest ally.\n\nDescription: {memory}", 'prompt_template': 'Example conversation:\n{prompt}<|eot_id|>', 'bot_template': '<|start_header_id|>{bot_name}<|end_header_id|>\n\n{message}<|eot_id|>', 'user_template': '<|start_header_id|>User<|end_header_id|>\n\n{message}<|eot_id|>', 'response_template': '<|start_header_id|>{bot_name}<|end_header_id|>\n\n', 'truncate_by_message': False}
generation_params: {'temperature': 1.0, 'top_p': 1.0, 'min_p': 0.0, 'top_k': 40, 'presence_penalty': 0.0, 'frequency_penalty': 0.0, 'stopping_words': ['<|eot_id|>', '\n'], 'max_input_tokens': 512, 'best_of': 16, 'max_output_tokens': 64}
is_internal_developer: False
language_model: anhnv125/Hydra-Llama-3
max_input_tokens: 512
max_output_tokens: 64
model_architecture: LlamaForCausalLM
model_eval_status: success
model_group: anhnv125/Hydra-Llama-3
model_name: anhnv125-hydra-llama-3_v4
model_num_parameters: 8030261248.0
model_repo: anhnv125/Hydra-Llama-3
model_size: 8B
num_battles: 6814
num_wins: 3607
ranking_group: single
reward_formatter: {'bot_template': 'Bot: {message}\n', 'memory_template': 'Memory: {memory}\n', 'prompt_template': '{prompt}\n', 'response_template': 'Bot:', 'truncate_by_message': False, 'user_template': 'User: {message}\n'}
reward_repo: rirv938/reward_gpt2_medium_preference_24m_e2
status: torndown
submission_type: basic
timestamp: 2024-04-21T15:24:42+00:00
us_pacific_date: 2024-04-21
win_ratio: 0.5293513354857646
Resubmit model
Running pipeline stage MKMLizer
Starting job with name anhnv125-hydra-llama-3-v4-mkmlizer
Waiting for job on anhnv125-hydra-llama-3-v4-mkmlizer to finish
anhnv125-hydra-llama-3-v4-mkmlizer: ╔═════════════════════════════════════════════════════════════════════╗
anhnv125-hydra-llama-3-v4-mkmlizer: ║ _____ __ __ ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ / _/ /_ ___ __/ / ___ ___ / / ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ / _/ / // / |/|/ / _ \/ -_) -_) / ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ /_//_/\_, /|__,__/_//_/\__/\__/_/ ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ /___/ ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ Version: 0.8.10 ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ Copyright 2023 MK ONE TECHNOLOGIES Inc. ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ The license key for the current software has been verified as ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ belonging to: ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ Chai Research Corp. ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ Account ID: 7997a29f-0ceb-4cc7-9adf-840c57b4ae6f ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ Expiration: 2024-07-15 23:59:59 ║
anhnv125-hydra-llama-3-v4-mkmlizer: ║ ║
anhnv125-hydra-llama-3-v4-mkmlizer: ╚═════════════════════════════════════════════════════════════════════╝
anhnv125-hydra-llama-3-v4-mkmlizer: /opt/conda/lib/python3.10/site-packages/huggingface_hub/utils/_deprecation.py:131: FutureWarning: 'list_files_info' (from 'huggingface_hub.hf_api') is deprecated and will be removed from version '0.23'. Use `list_repo_tree` and `get_paths_info` instead.
anhnv125-hydra-llama-3-v4-mkmlizer: warnings.warn(warning_message, FutureWarning)
anhnv125-hydra-llama-3-v4-mkmlizer: Downloaded to shared memory in 14.258s
anhnv125-hydra-llama-3-v4-mkmlizer: quantizing model to /dev/shm/model_cache
anhnv125-hydra-llama-3-v4-mkmlizer: Saving flywheel model at /dev/shm/model_cache
anhnv125-hydra-llama-3-v4-mkmlizer: Loading 0: 0%| | 0/291 [00:00<?, ?it/s] Loading 0: 49%|████▉ | 143/291 [00:01<00:01, 142.81it/s] Loading 0: 64%|██████▍ | 187/291 [00:06<00:04, 21.93it/s] Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.
anhnv125-hydra-llama-3-v4-mkmlizer: quantized model in 19.961s
anhnv125-hydra-llama-3-v4-mkmlizer: Processed model anhnv125/Hydra-Llama-3 in 35.226s
anhnv125-hydra-llama-3-v4-mkmlizer: creating bucket guanaco-mkml-models
anhnv125-hydra-llama-3-v4-mkmlizer: Bucket 's3://guanaco-mkml-models/' created
anhnv125-hydra-llama-3-v4-mkmlizer: uploading /dev/shm/model_cache to s3://guanaco-mkml-models/anhnv125-hydra-llama-3-v4
anhnv125-hydra-llama-3-v4-mkmlizer: cp /dev/shm/model_cache/config.json s3://guanaco-mkml-models/anhnv125-hydra-llama-3-v4/config.json
anhnv125-hydra-llama-3-v4-mkmlizer: cp /dev/shm/model_cache/special_tokens_map.json s3://guanaco-mkml-models/anhnv125-hydra-llama-3-v4/special_tokens_map.json
anhnv125-hydra-llama-3-v4-mkmlizer: cp /dev/shm/model_cache/tokenizer_config.json s3://guanaco-mkml-models/anhnv125-hydra-llama-3-v4/tokenizer_config.json
anhnv125-hydra-llama-3-v4-mkmlizer: cp /dev/shm/model_cache/tokenizer.json s3://guanaco-mkml-models/anhnv125-hydra-llama-3-v4/tokenizer.json
anhnv125-hydra-llama-3-v4-mkmlizer: cp /dev/shm/model_cache/flywheel_model.0.safetensors s3://guanaco-mkml-models/anhnv125-hydra-llama-3-v4/flywheel_model.0.safetensors
anhnv125-hydra-llama-3-v4-mkmlizer: loading reward model from rirv938/reward_gpt2_medium_preference_24m_e2
anhnv125-hydra-llama-3-v4-mkmlizer: /opt/conda/lib/python3.10/site-packages/transformers/models/auto/configuration_auto.py:913: FutureWarning: The `use_auth_token` argument is deprecated and will be removed in v5 of Transformers. Please use `token` instead.
anhnv125-hydra-llama-3-v4-mkmlizer: warnings.warn(
anhnv125-hydra-llama-3-v4-mkmlizer: /opt/conda/lib/python3.10/site-packages/transformers/models/auto/tokenization_auto.py:757: FutureWarning: The `use_auth_token` argument is deprecated and will be removed in v5 of Transformers. Please use `token` instead.
anhnv125-hydra-llama-3-v4-mkmlizer: warnings.warn(
anhnv125-hydra-llama-3-v4-mkmlizer: /opt/conda/lib/python3.10/site-packages/transformers/models/auto/auto_factory.py:468: FutureWarning: The `use_auth_token` argument is deprecated and will be removed in v5 of Transformers. Please use `token` instead.
anhnv125-hydra-llama-3-v4-mkmlizer: warnings.warn(
anhnv125-hydra-llama-3-v4-mkmlizer: /opt/conda/lib/python3.10/site-packages/torch/_utils.py:831: UserWarning: TypedStorage is deprecated. It will be removed in the future and UntypedStorage will be the only storage class. This should only matter to you if you are using storages directly. To access UntypedStorage directly, use tensor.untyped_storage() instead of tensor.storage()
anhnv125-hydra-llama-3-v4-mkmlizer: return self.fget.__get__(instance, owner)()
anhnv125-hydra-llama-3-v4-mkmlizer: Saving model to /tmp/reward_cache/reward.tensors
anhnv125-hydra-llama-3-v4-mkmlizer: Saving duration: 0.242s
anhnv125-hydra-llama-3-v4-mkmlizer: Processed model rirv938/reward_gpt2_medium_preference_24m_e2 in 4.867s
anhnv125-hydra-llama-3-v4-mkmlizer: creating bucket guanaco-reward-models
anhnv125-hydra-llama-3-v4-mkmlizer: Bucket 's3://guanaco-reward-models/' created
anhnv125-hydra-llama-3-v4-mkmlizer: uploading /tmp/reward_cache to s3://guanaco-reward-models/anhnv125-hydra-llama-3-v4_reward
anhnv125-hydra-llama-3-v4-mkmlizer: cp /tmp/reward_cache/special_tokens_map.json s3://guanaco-reward-models/anhnv125-hydra-llama-3-v4_reward/special_tokens_map.json
anhnv125-hydra-llama-3-v4-mkmlizer: cp /tmp/reward_cache/tokenizer_config.json s3://guanaco-reward-models/anhnv125-hydra-llama-3-v4_reward/tokenizer_config.json
anhnv125-hydra-llama-3-v4-mkmlizer: cp /tmp/reward_cache/config.json s3://guanaco-reward-models/anhnv125-hydra-llama-3-v4_reward/config.json
anhnv125-hydra-llama-3-v4-mkmlizer: cp /tmp/reward_cache/vocab.json s3://guanaco-reward-models/anhnv125-hydra-llama-3-v4_reward/vocab.json
anhnv125-hydra-llama-3-v4-mkmlizer: cp /tmp/reward_cache/merges.txt s3://guanaco-reward-models/anhnv125-hydra-llama-3-v4_reward/merges.txt
anhnv125-hydra-llama-3-v4-mkmlizer: cp /tmp/reward_cache/tokenizer.json s3://guanaco-reward-models/anhnv125-hydra-llama-3-v4_reward/tokenizer.json
anhnv125-hydra-llama-3-v4-mkmlizer: cp /tmp/reward_cache/reward.tensors s3://guanaco-reward-models/anhnv125-hydra-llama-3-v4_reward/reward.tensors
Job anhnv125-hydra-llama-3-v4-mkmlizer completed after 64.02s with status: succeeded
Stopping job with name anhnv125-hydra-llama-3-v4-mkmlizer
Pipeline stage MKMLizer completed in 69.32s
Running pipeline stage MKMLKubeTemplater
Pipeline stage MKMLKubeTemplater completed in 0.12s
Running pipeline stage ISVCDeployer
Creating inference service anhnv125-hydra-llama-3-v4
Waiting for inference service anhnv125-hydra-llama-3-v4 to be ready
Inference service anhnv125-hydra-llama-3-v4 ready after 40.21605134010315s
Pipeline stage ISVCDeployer completed in 48.25s
Running pipeline stage StressChecker
Received healthy response to inference request in 2.118086099624634s
Received healthy response to inference request in 1.3610191345214844s
Received healthy response to inference request in 1.3069534301757812s
Received healthy response to inference request in 1.2879343032836914s
Received healthy response to inference request in 1.2994318008422852s
5 requests
0 failed requests
5th percentile: 1.2902338027954101
10th percentile: 1.2925333023071288
20th percentile: 1.2971323013305665
30th percentile: 1.3009361267089843
40th percentile: 1.3039447784423828
50th percentile: 1.3069534301757812
60th percentile: 1.3285797119140625
70th percentile: 1.3502059936523438
80th percentile: 1.5124325275421144
90th percentile: 1.8152593135833741
95th percentile: 1.9666727066040037
99th percentile: 2.087803421020508
mean time: 1.4746849536895752
Pipeline stage StressChecker completed in 8.00s
Running pipeline stage DaemonicModelEvalScorer
Pipeline stage DaemonicModelEvalScorer completed in 0.03s
Running pipeline stage DaemonicSafetyScorer
Running M-Eval for topic stay_in_character
Pipeline stage DaemonicSafetyScorer completed in 0.05s
M-Eval Dataset for topic stay_in_character is loaded
anhnv125-hydra-llama-3_v4 status is now deployed due to DeploymentManager action
anhnv125-hydra-llama-3_v4 status is now inactive due to auto deactivation removed underperforming models
admin requested tearing down of anhnv125-hydra-llama-3_v4
Running pipeline stage ISVCDeleter
Checking if service anhnv125-hydra-llama-3-v4 is running
Tearing down inference service anhnv125-hydra-llama-3-v4
Toredown service anhnv125-hydra-llama-3-v4
Pipeline stage ISVCDeleter completed in 3.77s
Running pipeline stage MKMLModelDeleter
Cleaning model data from S3
Cleaning model data from model cache
Deleting key anhnv125-hydra-llama-3-v4/config.json from bucket guanaco-mkml-models
Deleting key anhnv125-hydra-llama-3-v4/flywheel_model.0.safetensors from bucket guanaco-mkml-models
Deleting key anhnv125-hydra-llama-3-v4/special_tokens_map.json from bucket guanaco-mkml-models
Deleting key anhnv125-hydra-llama-3-v4/tokenizer.json from bucket guanaco-mkml-models
Deleting key anhnv125-hydra-llama-3-v4/tokenizer_config.json from bucket guanaco-mkml-models
Cleaning model data from model cache
Deleting key anhnv125-hydra-llama-3-v4_reward/config.json from bucket guanaco-reward-models
Deleting key anhnv125-hydra-llama-3-v4_reward/merges.txt from bucket guanaco-reward-models
Deleting key anhnv125-hydra-llama-3-v4_reward/reward.tensors from bucket guanaco-reward-models
Deleting key anhnv125-hydra-llama-3-v4_reward/special_tokens_map.json from bucket guanaco-reward-models
Deleting key anhnv125-hydra-llama-3-v4_reward/tokenizer.json from bucket guanaco-reward-models
Deleting key anhnv125-hydra-llama-3-v4_reward/tokenizer_config.json from bucket guanaco-reward-models
Deleting key anhnv125-hydra-llama-3-v4_reward/vocab.json from bucket guanaco-reward-models
Pipeline stage MKMLModelDeleter completed in 2.85s
anhnv125-hydra-llama-3_v4 status is now torndown due to DeploymentManager action