submission_id: meseca-sophon-2_v1
developer_uid: nguyenzzz
status: inactive
model_repo: meseca/sophon-2
reward_repo: ChaiML/reward_gpt2_medium_preference_24m_e2
generation_params: {'temperature': 1.0, 'top_p': 1.0, 'min_p': 0.05, 'top_k': 40, 'presence_penalty': 0.0, 'frequency_penalty': 0.0, 'stopping_words': ['\n'], 'max_input_tokens': 512, 'best_of': 16, 'max_output_tokens': 64}
formatter: {'memory_template': "<|im_start|>system\n{bot_name}'s personality: {memory}\n", 'prompt_template': '{prompt}\n<START>\n', 'bot_template': '{bot_name}: {message}\n', 'user_template': '{user_name}: {message}\n', 'response_template': '{bot_name}:', 'truncate_by_message': True}
reward_formatter: {'memory_template': "{bot_name}'s Persona: {memory}\n####\n", 'prompt_template': '{prompt}\n<START>\n', 'bot_template': '{bot_name}: {message}\n', 'user_template': '{user_name}: {message}\n', 'response_template': '{bot_name}:', 'truncate_by_message': False}
timestamp: 2024-05-18T10:37:00+00:00
model_name: meseca-sophon-2_v1
model_eval_status: success
model_group: meseca/sophon-2
num_battles: 17128
num_wins: 9160
celo_rating: 1198.14
safety_score: 0.82
propriety_score: 0.0
propriety_total_count: 0.0
submission_type: basic
model_architecture: LlamaForCausalLM
model_num_parameters: 8030261248.0
best_of: 16
max_input_tokens: 512
max_output_tokens: 64
display_name: meseca-sophon-2_v1
ineligible_reason: propriety_total_count < 5000
language_model: meseca/sophon-2
model_size: 8B
reward_model: ChaiML/reward_gpt2_medium_preference_24m_e2
us_pacific_date: 2024-05-18
win_ratio: 0.5347968239140588
Resubmit model
Running pipeline stage MKMLizer
Starting job with name meseca-sophon-2-v1-mkmlizer
Waiting for job on meseca-sophon-2-v1-mkmlizer to finish
meseca-sophon-2-v1-mkmlizer: ╔═════════════════════════════════════════════════════════════════════╗
meseca-sophon-2-v1-mkmlizer: ║ _____ __ __ ║
meseca-sophon-2-v1-mkmlizer: ║ / _/ /_ ___ __/ / ___ ___ / / ║
meseca-sophon-2-v1-mkmlizer: ║ / _/ / // / |/|/ / _ \/ -_) -_) / ║
meseca-sophon-2-v1-mkmlizer: ║ /_//_/\_, /|__,__/_//_/\__/\__/_/ ║
meseca-sophon-2-v1-mkmlizer: ║ /___/ ║
meseca-sophon-2-v1-mkmlizer: ║ ║
meseca-sophon-2-v1-mkmlizer: ║ Version: 0.8.14 ║
meseca-sophon-2-v1-mkmlizer: ║ Copyright 2023 MK ONE TECHNOLOGIES Inc. ║
meseca-sophon-2-v1-mkmlizer: ║ https://mk1.ai ║
meseca-sophon-2-v1-mkmlizer: ║ ║
meseca-sophon-2-v1-mkmlizer: ║ The license key for the current software has been verified as ║
meseca-sophon-2-v1-mkmlizer: ║ belonging to: ║
meseca-sophon-2-v1-mkmlizer: ║ ║
meseca-sophon-2-v1-mkmlizer: ║ Chai Research Corp. ║
meseca-sophon-2-v1-mkmlizer: ║ Account ID: 7997a29f-0ceb-4cc7-9adf-840c57b4ae6f ║
meseca-sophon-2-v1-mkmlizer: ║ Expiration: 2024-07-15 23:59:59 ║
meseca-sophon-2-v1-mkmlizer: ║ ║
meseca-sophon-2-v1-mkmlizer: ╚═════════════════════════════════════════════════════════════════════╝
meseca-sophon-2-v1-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.
meseca-sophon-2-v1-mkmlizer: warnings.warn(warning_message, FutureWarning)
meseca-sophon-2-v1-mkmlizer: Downloaded to shared memory in 27.095s
meseca-sophon-2-v1-mkmlizer: quantizing model to /dev/shm/model_cache
meseca-sophon-2-v1-mkmlizer: Saving flywheel model at /dev/shm/model_cache
meseca-sophon-2-v1-mkmlizer: Loading 0: 0%| | 0/291 [00:00<?, ?it/s] Loading 0: 5%|▌ | 15/291 [00:00<00:01, 149.04it/s] Loading 0: 12%|█▏ | 35/291 [00:00<00:01, 178.39it/s] Loading 0: 20%|█▉ | 57/291 [00:00<00:01, 191.66it/s] Loading 0: 26%|██▋ | 77/291 [00:00<00:01, 184.48it/s] Loading 0: 33%|███▎ | 96/291 [00:00<00:02, 94.27it/s] Loading 0: 41%|████ | 120/291 [00:00<00:01, 121.77it/s] Loading 0: 48%|████▊ | 139/291 [00:01<00:01, 127.57it/s] Loading 0: 54%|█████▍ | 158/291 [00:01<00:00, 138.41it/s] Loading 0: 62%|██████▏ | 181/291 [00:01<00:00, 156.98it/s] Loading 0: 68%|██████▊ | 199/291 [00:01<00:00, 97.13it/s] Loading 0: 75%|███████▌ | 219/291 [00:01<00:00, 114.59it/s] Loading 0: 82%|████████▏ | 238/291 [00:01<00:00, 127.92it/s] Loading 0: 88%|████████▊ | 257/291 [00:01<00:00, 141.52it/s] Loading 0: 96%|█████████▌| 279/291 [00:02<00:00, 159.36it/s] Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.
meseca-sophon-2-v1-mkmlizer: quantized model in 17.531s
meseca-sophon-2-v1-mkmlizer: Processed model meseca/sophon-2 in 45.644s
meseca-sophon-2-v1-mkmlizer: creating bucket guanaco-mkml-models
meseca-sophon-2-v1-mkmlizer: Bucket 's3://guanaco-mkml-models/' created
meseca-sophon-2-v1-mkmlizer: uploading /dev/shm/model_cache to s3://guanaco-mkml-models/meseca-sophon-2-v1
meseca-sophon-2-v1-mkmlizer: cp /dev/shm/model_cache/tokenizer_config.json s3://guanaco-mkml-models/meseca-sophon-2-v1/tokenizer_config.json
meseca-sophon-2-v1-mkmlizer: cp /dev/shm/model_cache/config.json s3://guanaco-mkml-models/meseca-sophon-2-v1/config.json
meseca-sophon-2-v1-mkmlizer: cp /dev/shm/model_cache/special_tokens_map.json s3://guanaco-mkml-models/meseca-sophon-2-v1/special_tokens_map.json
meseca-sophon-2-v1-mkmlizer: cp /dev/shm/model_cache/tokenizer.json s3://guanaco-mkml-models/meseca-sophon-2-v1/tokenizer.json
meseca-sophon-2-v1-mkmlizer: cp /dev/shm/model_cache/flywheel_model.0.safetensors s3://guanaco-mkml-models/meseca-sophon-2-v1/flywheel_model.0.safetensors
meseca-sophon-2-v1-mkmlizer: loading reward model from ChaiML/reward_gpt2_medium_preference_24m_e2
meseca-sophon-2-v1-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.
meseca-sophon-2-v1-mkmlizer: warnings.warn(
meseca-sophon-2-v1-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.
meseca-sophon-2-v1-mkmlizer: warnings.warn(
meseca-sophon-2-v1-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.
meseca-sophon-2-v1-mkmlizer: warnings.warn(
meseca-sophon-2-v1-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()
meseca-sophon-2-v1-mkmlizer: return self.fget.__get__(instance, owner)()
meseca-sophon-2-v1-mkmlizer: Saving model to /tmp/reward_cache/reward.tensors
meseca-sophon-2-v1-mkmlizer: Saving duration: 0.227s
meseca-sophon-2-v1-mkmlizer: Processed model ChaiML/reward_gpt2_medium_preference_24m_e2 in 3.574s
meseca-sophon-2-v1-mkmlizer: creating bucket guanaco-reward-models
meseca-sophon-2-v1-mkmlizer: Bucket 's3://guanaco-reward-models/' created
meseca-sophon-2-v1-mkmlizer: uploading /tmp/reward_cache to s3://guanaco-reward-models/meseca-sophon-2-v1_reward
meseca-sophon-2-v1-mkmlizer: cp /tmp/reward_cache/config.json s3://guanaco-reward-models/meseca-sophon-2-v1_reward/config.json
meseca-sophon-2-v1-mkmlizer: cp /tmp/reward_cache/special_tokens_map.json s3://guanaco-reward-models/meseca-sophon-2-v1_reward/special_tokens_map.json
meseca-sophon-2-v1-mkmlizer: cp /tmp/reward_cache/merges.txt s3://guanaco-reward-models/meseca-sophon-2-v1_reward/merges.txt
meseca-sophon-2-v1-mkmlizer: cp /tmp/reward_cache/tokenizer_config.json s3://guanaco-reward-models/meseca-sophon-2-v1_reward/tokenizer_config.json
meseca-sophon-2-v1-mkmlizer: cp /tmp/reward_cache/vocab.json s3://guanaco-reward-models/meseca-sophon-2-v1_reward/vocab.json
meseca-sophon-2-v1-mkmlizer: cp /tmp/reward_cache/tokenizer.json s3://guanaco-reward-models/meseca-sophon-2-v1_reward/tokenizer.json
meseca-sophon-2-v1-mkmlizer: cp /tmp/reward_cache/reward.tensors s3://guanaco-reward-models/meseca-sophon-2-v1_reward/reward.tensors
Job meseca-sophon-2-v1-mkmlizer completed after 73.18s with status: succeeded
Stopping job with name meseca-sophon-2-v1-mkmlizer
Pipeline stage MKMLizer completed in 76.28s
Running pipeline stage MKMLKubeTemplater
Pipeline stage MKMLKubeTemplater completed in 0.10s
Running pipeline stage ISVCDeployer
Creating inference service meseca-sophon-2-v1
Waiting for inference service meseca-sophon-2-v1 to be ready
Inference service meseca-sophon-2-v1 ready after 41.43997502326965s
Pipeline stage ISVCDeployer completed in 48.41s
Running pipeline stage StressChecker
Received healthy response to inference request in 2.154219150543213s
Received healthy response to inference request in 1.112442970275879s
Received healthy response to inference request in 1.226715087890625s
Received healthy response to inference request in 1.2272088527679443s
Received healthy response to inference request in 1.0377097129821777s
5 requests
0 failed requests
5th percentile: 1.052656364440918
10th percentile: 1.0676030158996581
20th percentile: 1.0974963188171387
30th percentile: 1.1352973937988282
40th percentile: 1.1810062408447266
50th percentile: 1.226715087890625
60th percentile: 1.2269125938415528
70th percentile: 1.2271100997924804
80th percentile: 1.4126109123229982
90th percentile: 1.7834150314331056
95th percentile: 1.968817090988159
99th percentile: 2.1171387386322023
mean time: 1.3516591548919679
Pipeline stage StressChecker completed in 7.40s
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 0.04s
M-Eval Dataset for topic stay_in_character is loaded
meseca-sophon-2_v1 status is now deployed due to DeploymentManager action
meseca-sophon-2_v1 status is now inactive due to auto deactivation removed underperforming models

Usage Metrics

Latency Metrics