submission_id: steelskull-l3-arcania-4x8b_v1
developer_uid: Steelskull
status: inactive
model_repo: Steelskull/L3-Arcania-4x8b
reward_repo: ChaiML/reward_gpt2_medium_preference_24m_e2
generation_params: {'temperature': 0.85, 'top_p': 1.0, 'min_p': 0.05, 'top_k': 90, 'presence_penalty': 0.0, 'frequency_penalty': 0.0, 'stopping_words': ['\n'], 'max_input_tokens': 512, 'best_of': 4, 'max_output_tokens': 64}
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}
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: 2024-05-19T21:54:09+00:00
model_name: steelskull-l3-arcania-4x8b_v1
model_eval_status: success
model_group: Steelskull/L3-Arcania-4x
num_battles: 17026
num_wins: 8905
celo_rating: 1179.85
safety_score: 0.97
propriety_score: 0.0
propriety_total_count: 0.0
submission_type: basic
model_architecture: MixtralForCausalLM
model_num_parameters: 24942219264.0
best_of: 4
max_input_tokens: 512
max_output_tokens: 64
display_name: steelskull-l3-arcania-4x8b_v1
ineligible_reason: propriety_total_count < 5000
language_model: Steelskull/L3-Arcania-4x8b
model_size: 25B
reward_model: ChaiML/reward_gpt2_medium_preference_24m_e2
us_pacific_date: 2024-05-19
win_ratio: 0.5230236109479619
Resubmit model
Running pipeline stage MKMLizer
Starting job with name steelskull-l3-arcania-4x8b-v1-mkmlizer
Waiting for job on steelskull-l3-arcania-4x8b-v1-mkmlizer to finish
steelskull-l3-arcania-4x8b-v1-mkmlizer: ╔═════════════════════════════════════════════════════════════════════╗
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ _____ __ __ ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ / _/ /_ ___ __/ / ___ ___ / / ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ / _/ / // / |/|/ / _ \/ -_) -_) / ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ /_//_/\_, /|__,__/_//_/\__/\__/_/ ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ /___/ ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ Version: 0.8.14 ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ Copyright 2023 MK ONE TECHNOLOGIES Inc. ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ https://mk1.ai ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ The license key for the current software has been verified as ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ belonging to: ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ Chai Research Corp. ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ Account ID: 7997a29f-0ceb-4cc7-9adf-840c57b4ae6f ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ Expiration: 2024-07-15 23:59:59 ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ║ ║
steelskull-l3-arcania-4x8b-v1-mkmlizer: ╚═════════════════════════════════════════════════════════════════════╝
steelskull-l3-arcania-4x8b-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.
steelskull-l3-arcania-4x8b-v1-mkmlizer: warnings.warn(warning_message, FutureWarning)
steelskull-l3-arcania-4x8b-v1-mkmlizer: Downloaded to shared memory in 51.116s
steelskull-l3-arcania-4x8b-v1-mkmlizer: quantizing model to /dev/shm/model_cache
steelskull-l3-arcania-4x8b-v1-mkmlizer: Saving flywheel model at /dev/shm/model_cache
steelskull-l3-arcania-4x8b-v1-mkmlizer: Loading 0: 0%| | 0/611 [00:00<?, ?it/s] Loading 0: 0%| | 2/611 [00:04<21:05, 2.08s/it] Loading 0: 2%|▏ | 13/611 [00:04<02:25, 4.11it/s] Loading 0: 4%|▍ | 24/611 [00:04<01:06, 8.89it/s] Loading 0: 6%|▌ | 35/611 [00:04<00:38, 15.01it/s] Loading 0: 7%|▋ | 45/611 [00:04<00:26, 21.56it/s] Loading 0: 11%|█ | 68/611 [00:04<00:12, 42.48it/s] Loading 0: 13%|█▎ | 82/611 [00:04<00:09, 54.40it/s] Loading 0: 16%|█▌ | 96/611 [00:04<00:08, 63.65it/s] Loading 0: 18%|█▊ | 109/611 [00:05<00:12, 39.75it/s] Loading 0: 19%|█▉ | 119/611 [00:05<00:10, 45.94it/s] Loading 0: 21%|██ | 129/611 [00:05<00:09, 52.86it/s] Loading 0: 23%|██▎ | 139/611 [00:05<00:07, 59.83it/s] Loading 0: 24%|██▍ | 149/611 [00:05<00:06, 66.05it/s] Loading 0: 26%|██▌ | 159/611 [00:06<00:06, 71.92it/s] Loading 0: 28%|██▊ | 169/611 [00:06<00:05, 76.70it/s] Loading 0: 29%|██▉ | 179/611 [00:06<00:05, 80.71it/s] Loading 0: 31%|███ | 189/611 [00:06<00:10, 39.29it/s] Loading 0: 32%|███▏ | 198/611 [00:06<00:08, 46.43it/s] Loading 0: 34%|███▍ | 208/611 [00:07<00:07, 54.73it/s] Loading 0: 36%|███▌ | 218/611 [00:07<00:06, 62.35it/s] Loading 0: 37%|███▋ | 227/611 [00:07<00:05, 67.46it/s] Loading 0: 39%|███▊ | 236/611 [00:07<00:05, 72.54it/s] Loading 0: 40%|████ | 245/611 [00:17<01:58, 3.09it/s] Loading 0: 42%|████▏ | 255/611 [00:17<01:20, 4.45it/s] Loading 0: 43%|████▎ | 265/611 [00:17<00:54, 6.31it/s] Loading 0: 45%|████▍ | 273/611 [00:17<00:45, 7.37it/s] Loading 0: 46%|████▋ | 283/611 [00:18<00:31, 10.42it/s] Loading 0: 48%|████▊ | 292/611 [00:18<00:22, 14.01it/s] Loading 0: 49%|████▉ | 302/611 [00:18<00:16, 19.16it/s] Loading 0: 51%|█████ | 312/611 [00:18<00:11, 25.37it/s] Loading 0: 53%|█████▎ | 321/611 [00:18<00:09, 31.76it/s] Loading 0: 54%|█████▍ | 331/611 [00:18<00:07, 39.83it/s] Loading 0: 56%|█████▌ | 341/611 [00:18<00:05, 48.23it/s] Loading 0: 57%|█████▋ | 351/611 [00:18<00:04, 56.48it/s] Loading 0: 59%|█████▉ | 360/611 [00:19<00:07, 34.25it/s] Loading 0: 68%|██████▊ | 414/611 [00:19<00:01, 100.55it/s] Loading 0: 71%|███████ | 434/611 [00:19<00:01, 116.05it/s] Loading 0: 75%|███████▍ | 458/611 [00:19<00:01, 138.82it/s] Loading 0: 79%|███████▊ | 480/611 [00:19<00:00, 137.87it/s] Loading 0: 82%|████████▏ | 500/611 [00:19<00:00, 149.39it/s] Loading 0: 85%|████████▌ | 520/611 [00:19<00:00, 160.07it/s] Loading 0: 88%|████████▊ | 540/611 [00:20<00:00, 167.95it/s] Loading 0: 92%|█████████▏| 564/611 [00:20<00:00, 186.24it/s] Loading 0: 99%|█████████▉| 605/611 [00:20<00:00, 245.77it/s] Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.
steelskull-l3-arcania-4x8b-v1-mkmlizer: quantized model in 34.219s
steelskull-l3-arcania-4x8b-v1-mkmlizer: Processed model Steelskull/L3-Arcania-4x8b in 88.235s
steelskull-l3-arcania-4x8b-v1-mkmlizer: Bucket 's3://guanaco-mkml-models/' created
steelskull-l3-arcania-4x8b-v1-mkmlizer: uploading /dev/shm/model_cache to s3://guanaco-mkml-models/steelskull-l3-arcania-4x8b-v1
steelskull-l3-arcania-4x8b-v1-mkmlizer: cp /dev/shm/model_cache/special_tokens_map.json s3://guanaco-mkml-models/steelskull-l3-arcania-4x8b-v1/special_tokens_map.json
steelskull-l3-arcania-4x8b-v1-mkmlizer: cp /dev/shm/model_cache/config.json s3://guanaco-mkml-models/steelskull-l3-arcania-4x8b-v1/config.json
steelskull-l3-arcania-4x8b-v1-mkmlizer: cp /dev/shm/model_cache/tokenizer_config.json s3://guanaco-mkml-models/steelskull-l3-arcania-4x8b-v1/tokenizer_config.json
steelskull-l3-arcania-4x8b-v1-mkmlizer: cp /dev/shm/model_cache/tokenizer.json s3://guanaco-mkml-models/steelskull-l3-arcania-4x8b-v1/tokenizer.json
steelskull-l3-arcania-4x8b-v1-mkmlizer: cp /dev/shm/model_cache/flywheel_model.0.safetensors s3://guanaco-mkml-models/steelskull-l3-arcania-4x8b-v1/flywheel_model.0.safetensors
steelskull-l3-arcania-4x8b-v1-mkmlizer: loading reward model from ChaiML/reward_gpt2_medium_preference_24m_e2
steelskull-l3-arcania-4x8b-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.
steelskull-l3-arcania-4x8b-v1-mkmlizer: warnings.warn(
steelskull-l3-arcania-4x8b-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.
steelskull-l3-arcania-4x8b-v1-mkmlizer: warnings.warn(
steelskull-l3-arcania-4x8b-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.
steelskull-l3-arcania-4x8b-v1-mkmlizer: warnings.warn(
steelskull-l3-arcania-4x8b-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()
steelskull-l3-arcania-4x8b-v1-mkmlizer: return self.fget.__get__(instance, owner)()
steelskull-l3-arcania-4x8b-v1-mkmlizer: Saving model to /tmp/reward_cache/reward.tensors
steelskull-l3-arcania-4x8b-v1-mkmlizer: Saving duration: 0.223s
steelskull-l3-arcania-4x8b-v1-mkmlizer: Processed model ChaiML/reward_gpt2_medium_preference_24m_e2 in 4.619s
steelskull-l3-arcania-4x8b-v1-mkmlizer: creating bucket guanaco-reward-models
steelskull-l3-arcania-4x8b-v1-mkmlizer: Bucket 's3://guanaco-reward-models/' created
steelskull-l3-arcania-4x8b-v1-mkmlizer: uploading /tmp/reward_cache to s3://guanaco-reward-models/steelskull-l3-arcania-4x8b-v1_reward
steelskull-l3-arcania-4x8b-v1-mkmlizer: cp /tmp/reward_cache/config.json s3://guanaco-reward-models/steelskull-l3-arcania-4x8b-v1_reward/config.json
steelskull-l3-arcania-4x8b-v1-mkmlizer: cp /tmp/reward_cache/tokenizer_config.json s3://guanaco-reward-models/steelskull-l3-arcania-4x8b-v1_reward/tokenizer_config.json
steelskull-l3-arcania-4x8b-v1-mkmlizer: cp /tmp/reward_cache/merges.txt s3://guanaco-reward-models/steelskull-l3-arcania-4x8b-v1_reward/merges.txt
steelskull-l3-arcania-4x8b-v1-mkmlizer: cp /tmp/reward_cache/vocab.json s3://guanaco-reward-models/steelskull-l3-arcania-4x8b-v1_reward/vocab.json
steelskull-l3-arcania-4x8b-v1-mkmlizer: cp /tmp/reward_cache/tokenizer.json s3://guanaco-reward-models/steelskull-l3-arcania-4x8b-v1_reward/tokenizer.json
steelskull-l3-arcania-4x8b-v1-mkmlizer: cp /tmp/reward_cache/special_tokens_map.json s3://guanaco-reward-models/steelskull-l3-arcania-4x8b-v1_reward/special_tokens_map.json
steelskull-l3-arcania-4x8b-v1-mkmlizer: cp /tmp/reward_cache/reward.tensors s3://guanaco-reward-models/steelskull-l3-arcania-4x8b-v1_reward/reward.tensors
Job steelskull-l3-arcania-4x8b-v1-mkmlizer completed after 126.73s with status: succeeded
Stopping job with name steelskull-l3-arcania-4x8b-v1-mkmlizer
Pipeline stage MKMLizer completed in 129.35s
Running pipeline stage MKMLKubeTemplater
Pipeline stage MKMLKubeTemplater completed in 0.10s
Running pipeline stage ISVCDeployer
Creating inference service steelskull-l3-arcania-4x8b-v1
Waiting for inference service steelskull-l3-arcania-4x8b-v1 to be ready
Inference service steelskull-l3-arcania-4x8b-v1 ready after 40.27813005447388s
Pipeline stage ISVCDeployer completed in 47.21s
Running pipeline stage StressChecker
Received healthy response to inference request in 3.0738067626953125s
Received healthy response to inference request in 2.4090323448181152s
Received healthy response to inference request in 2.4072265625s
Received healthy response to inference request in 2.398010492324829s
Received healthy response to inference request in 2.2676961421966553s
5 requests
0 failed requests
5th percentile: 2.29375901222229
10th percentile: 2.3198218822479246
20th percentile: 2.3719476222991944
30th percentile: 2.3998537063598633
40th percentile: 2.4035401344299316
50th percentile: 2.4072265625
60th percentile: 2.407948875427246
70th percentile: 2.408671188354492
80th percentile: 2.541987228393555
90th percentile: 2.8078969955444335
95th percentile: 2.940851879119873
99th percentile: 3.0472157859802245
mean time: 2.5111544609069822
Pipeline stage StressChecker completed in 13.20s
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 0.03s
M-Eval Dataset for topic stay_in_character is loaded
steelskull-l3-arcania-4x8b_v1 status is now deployed due to DeploymentManager action
steelskull-l3-arcania-4x8b_v1 status is now inactive due to auto deactivation removed underperforming models

Usage Metrics

Latency Metrics