submission_id: hastagaras-llama-3-8b-ok_5775_v3
developer_uid: Hastagaras
status: inactive
model_repo: Hastagaras/llama-3-8b-okay-two
reward_repo: ChaiML/reward_gpt2_medium_preference_24m_e2
generation_params: {'temperature': 0.9, 'top_p': 0.9, 'min_p': 0.0, 'top_k': 40, 'presence_penalty': 0.0, 'frequency_penalty': 1.1, 'stopping_words': ['\n', '<|eot_id|>'], 'max_input_tokens': 512, 'best_of': 16, 'max_output_tokens': 64}
formatter: {'memory_template': "<|start_header_id|>system<|end_header_id|>\n\nAlways write your response as {bot_name} based on the following persona.\n\n{bot_name}'s Persona: {memory}\n", 'prompt_template': 'Previously: {prompt}<|eot_id|>', 'bot_template': '<|start_header_id|>assistant<|end_header_id|>\n\n{bot_name}: {message}<|eot_id|>', 'user_template': '<|start_header_id|>user<|end_header_id|>\n\n{user_name}: {message}<|eot_id|>', 'response_template': '<|start_header_id|>assistant<|end_header_id|>\n\n{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: {message}\n', 'user_template': 'User: {message}\n', 'response_template': 'Bot:', 'truncate_by_message': True}
timestamp: 2024-05-18T11:23:30+00:00
model_name: okay-two-is-okay
model_eval_status: success
model_group: Hastagaras/llama-3-8b-ok
num_battles: 21077
num_wins: 11129
celo_rating: 1197.6
safety_score: 0.96
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: okay-two-is-okay
ineligible_reason: propriety_total_count < 5000
language_model: Hastagaras/llama-3-8b-okay-two
model_size: 8B
reward_model: ChaiML/reward_gpt2_medium_preference_24m_e2
us_pacific_date: 2024-05-18
win_ratio: 0.5280163211083171
Resubmit model
Running pipeline stage MKMLizer
Starting job with name hastagaras-llama-3-8b-ok-5775-v3-mkmlizer
Waiting for job on hastagaras-llama-3-8b-ok-5775-v3-mkmlizer to finish
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ╔═════════════════════════════════════════════════════════════════════╗
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ _____ __ __ ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ / _/ /_ ___ __/ / ___ ___ / / ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ / _/ / // / |/|/ / _ \/ -_) -_) / ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ /_//_/\_, /|__,__/_//_/\__/\__/_/ ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ /___/ ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ Version: 0.8.14 ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ Copyright 2023 MK ONE TECHNOLOGIES Inc. ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ https://mk1.ai ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ The license key for the current software has been verified as ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ belonging to: ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ Chai Research Corp. ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ Account ID: 7997a29f-0ceb-4cc7-9adf-840c57b4ae6f ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ Expiration: 2024-07-15 23:59:59 ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ║ ║
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: ╚═════════════════════════════════════════════════════════════════════╝
hastagaras-llama-3-8b-ok-5775-v3-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.
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: warnings.warn(warning_message, FutureWarning)
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: Downloaded to shared memory in 11.867s
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: quantizing model to /dev/shm/model_cache
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: Saving flywheel model at /dev/shm/model_cache
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: Loading 0: 0%| | 0/291 [00:00<?, ?it/s] Loading 0: 7%|▋ | 21/291 [00:00<00:01, 196.76it/s] Loading 0: 14%|█▍ | 41/291 [00:00<00:01, 183.02it/s] Loading 0: 22%|██▏ | 63/291 [00:00<00:01, 196.01it/s] Loading 0: 29%|██▊ | 83/291 [00:00<00:02, 83.97it/s] Loading 0: 35%|███▌ | 102/291 [00:00<00:01, 103.13it/s] Loading 0: 42%|████▏ | 121/291 [00:01<00:01, 116.50it/s] Loading 0: 48%|████▊ | 140/291 [00:01<00:01, 130.53it/s] Loading 0: 57%|█████▋ | 165/291 [00:01<00:00, 155.20it/s] Loading 0: 64%|██████▍ | 187/291 [00:01<00:01, 90.62it/s] Loading 0: 70%|██████▉ | 203/291 [00:01<00:00, 100.42it/s] Loading 0: 78%|███████▊ | 228/291 [00:01<00:00, 125.81it/s] Loading 0: 85%|████████▍ | 247/291 [00:02<00:00, 137.43it/s] Loading 0: 92%|█████████▏| 268/291 [00:02<00:00, 153.50it/s] Loading 0: 99%|█████████▊| 287/291 [00:07<00:00, 12.62it/s] Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: quantized model in 17.655s
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: Processed model Hastagaras/llama-3-8b-okay-two in 30.575s
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: creating bucket guanaco-mkml-models
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: Bucket 's3://guanaco-mkml-models/' created
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: uploading /dev/shm/model_cache to s3://guanaco-mkml-models/hastagaras-llama-3-8b-ok-5775-v3
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: cp /dev/shm/model_cache/tokenizer_config.json s3://guanaco-mkml-models/hastagaras-llama-3-8b-ok-5775-v3/tokenizer_config.json
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: cp /dev/shm/model_cache/special_tokens_map.json s3://guanaco-mkml-models/hastagaras-llama-3-8b-ok-5775-v3/special_tokens_map.json
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: cp /dev/shm/model_cache/config.json s3://guanaco-mkml-models/hastagaras-llama-3-8b-ok-5775-v3/config.json
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: cp /dev/shm/model_cache/tokenizer.json s3://guanaco-mkml-models/hastagaras-llama-3-8b-ok-5775-v3/tokenizer.json
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: cp /dev/shm/model_cache/flywheel_model.0.safetensors s3://guanaco-mkml-models/hastagaras-llama-3-8b-ok-5775-v3/flywheel_model.0.safetensors
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: loading reward model from ChaiML/reward_gpt2_medium_preference_24m_e2
hastagaras-llama-3-8b-ok-5775-v3-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.
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: warnings.warn(
hastagaras-llama-3-8b-ok-5775-v3-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.
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: warnings.warn(
hastagaras-llama-3-8b-ok-5775-v3-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.
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: warnings.warn(
hastagaras-llama-3-8b-ok-5775-v3-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()
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: return self.fget.__get__(instance, owner)()
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: Saving model to /tmp/reward_cache/reward.tensors
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: Saving duration: 0.251s
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: Processed model ChaiML/reward_gpt2_medium_preference_24m_e2 in 3.334s
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: creating bucket guanaco-reward-models
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: Bucket 's3://guanaco-reward-models/' created
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: uploading /tmp/reward_cache to s3://guanaco-reward-models/hastagaras-llama-3-8b-ok-5775-v3_reward
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: cp /tmp/reward_cache/config.json s3://guanaco-reward-models/hastagaras-llama-3-8b-ok-5775-v3_reward/config.json
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: cp /tmp/reward_cache/tokenizer_config.json s3://guanaco-reward-models/hastagaras-llama-3-8b-ok-5775-v3_reward/tokenizer_config.json
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: cp /tmp/reward_cache/special_tokens_map.json s3://guanaco-reward-models/hastagaras-llama-3-8b-ok-5775-v3_reward/special_tokens_map.json
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: cp /tmp/reward_cache/vocab.json s3://guanaco-reward-models/hastagaras-llama-3-8b-ok-5775-v3_reward/vocab.json
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: cp /tmp/reward_cache/merges.txt s3://guanaco-reward-models/hastagaras-llama-3-8b-ok-5775-v3_reward/merges.txt
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: cp /tmp/reward_cache/tokenizer.json s3://guanaco-reward-models/hastagaras-llama-3-8b-ok-5775-v3_reward/tokenizer.json
hastagaras-llama-3-8b-ok-5775-v3-mkmlizer: cp /tmp/reward_cache/reward.tensors s3://guanaco-reward-models/hastagaras-llama-3-8b-ok-5775-v3_reward/reward.tensors
Job hastagaras-llama-3-8b-ok-5775-v3-mkmlizer completed after 52.63s with status: succeeded
Stopping job with name hastagaras-llama-3-8b-ok-5775-v3-mkmlizer
Pipeline stage MKMLizer completed in 55.64s
Running pipeline stage MKMLKubeTemplater
Pipeline stage MKMLKubeTemplater completed in 0.11s
Running pipeline stage ISVCDeployer
Creating inference service hastagaras-llama-3-8b-ok-5775-v3
Waiting for inference service hastagaras-llama-3-8b-ok-5775-v3 to be ready
Inference service hastagaras-llama-3-8b-ok-5775-v3 ready after 30.29911470413208s
Pipeline stage ISVCDeployer completed in 38.01s
Running pipeline stage StressChecker
Received healthy response to inference request in 2.078219413757324s
Received healthy response to inference request in 1.2838671207427979s
Received healthy response to inference request in 1.2480149269104004s
Received healthy response to inference request in 1.306687831878662s
Received healthy response to inference request in 1.3012845516204834s
5 requests
0 failed requests
5th percentile: 1.2551853656768799
10th percentile: 1.2623558044433594
20th percentile: 1.2766966819763184
30th percentile: 1.287350606918335
40th percentile: 1.2943175792694093
50th percentile: 1.3012845516204834
60th percentile: 1.303445863723755
70th percentile: 1.3056071758270265
80th percentile: 1.4609941482543947
90th percentile: 1.7696067810058596
95th percentile: 1.9239130973815917
99th percentile: 2.047358150482178
mean time: 1.4436147689819336
Pipeline stage StressChecker completed in 7.90s
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
hastagaras-llama-3-8b-ok_5775_v3 status is now deployed due to DeploymentManager action
hastagaras-llama-3-8b-ok_5775_v3 status is now inactive due to auto deactivation removed underperforming models

Usage Metrics

Latency Metrics