submission_id: dreamgen-opus-v1-2-llama-3-8b_v4
developer_uid: NuckaNon
status: torndown
model_repo: dreamgen/opus-v1.2-llama-3-8b
reward_repo: ChaiML/reward_gpt2_medium_preference_24m_e2
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': ['\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: 2024-04-20T11:25:24+00:00
model_name: dreamgen-opus-v1-2-llama-3-8b_v4
model_eval_status: success
model_group: dreamgen/opus-v1.2-llama
num_battles: 5347
num_wins: 2559
celo_rating: 1132.64
propriety_score: 0.0
propriety_total_count: 0.0
submission_type: basic
model_architecture: LlamaForCausalLM
model_num_parameters: 8030261248.0
best_of: 4
max_input_tokens: 512
max_output_tokens: 64
display_name: dreamgen-opus-v1-2-llama-3-8b_v4
ineligible_reason: propriety_total_count < 800
language_model: dreamgen/opus-v1.2-llama-3-8b
model_size: 8B
reward_model: ChaiML/reward_gpt2_medium_preference_24m_e2
us_pacific_date: 2024-04-20
win_ratio: 0.47858612305965964
preference_data_url: None
Resubmit model
Running pipeline stage MKMLizer
Starting job with name dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer
Waiting for job on dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer to finish
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ╔═════════════════════════════════════════════════════════════════════╗
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ _____ __ __ ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ / _/ /_ ___ __/ / ___ ___ / / ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ / _/ / // / |/|/ / _ \/ -_) -_) / ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ /_//_/\_, /|__,__/_//_/\__/\__/_/ ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ /___/ ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ Version: 0.8.10 ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ Copyright 2023 MK ONE TECHNOLOGIES Inc. ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ The license key for the current software has been verified as ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ belonging to: ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ Chai Research Corp. ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ Account ID: 7997a29f-0ceb-4cc7-9adf-840c57b4ae6f ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ Expiration: 2024-07-15 23:59:59 ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ║ ║
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: ╚═════════════════════════════════════════════════════════════════════╝
dreamgen-opus-v1-2-llama-3-8b-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.
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: warnings.warn(warning_message, FutureWarning)
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: Downloaded to shared memory in 22.056s
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: quantizing model to /dev/shm/model_cache
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: Saving flywheel model at /dev/shm/model_cache
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: Loading 0: 0%| | 0/291 [00:00<?, ?it/s] Loading 0: 35%|███▌ | 103/291 [00:01<00:01, 101.61it/s] Loading 0: 70%|██████▉ | 203/291 [00:02<00:00, 100.04it/s] Loading 0: 99%|█████████▊| 287/291 [00:07<00:00, 28.78it/s] Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: quantized model in 19.173s
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: Processed model dreamgen/opus-v1.2-llama-3-8b in 42.215s
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: creating bucket guanaco-mkml-models
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: Bucket 's3://guanaco-mkml-models/' created
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: uploading /dev/shm/model_cache to s3://guanaco-mkml-models/dreamgen-opus-v1-2-llama-3-8b-v4
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: cp /dev/shm/model_cache/tokenizer_config.json s3://guanaco-mkml-models/dreamgen-opus-v1-2-llama-3-8b-v4/tokenizer_config.json
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: cp /dev/shm/model_cache/special_tokens_map.json s3://guanaco-mkml-models/dreamgen-opus-v1-2-llama-3-8b-v4/special_tokens_map.json
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: cp /dev/shm/model_cache/config.json s3://guanaco-mkml-models/dreamgen-opus-v1-2-llama-3-8b-v4/config.json
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: cp /dev/shm/model_cache/tokenizer.json s3://guanaco-mkml-models/dreamgen-opus-v1-2-llama-3-8b-v4/tokenizer.json
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: cp /dev/shm/model_cache/flywheel_model.0.safetensors s3://guanaco-mkml-models/dreamgen-opus-v1-2-llama-3-8b-v4/flywheel_model.0.safetensors
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: loading reward model from ChaiML/reward_gpt2_medium_preference_24m_e2
dreamgen-opus-v1-2-llama-3-8b-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.
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: warnings.warn(
dreamgen-opus-v1-2-llama-3-8b-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.
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: warnings.warn(
dreamgen-opus-v1-2-llama-3-8b-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.
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: warnings.warn(
dreamgen-opus-v1-2-llama-3-8b-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()
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: return self.fget.__get__(instance, owner)()
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: Saving model to /tmp/reward_cache/reward.tensors
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: Saving duration: 0.223s
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: Processed model ChaiML/reward_gpt2_medium_preference_24m_e2 in 4.715s
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: creating bucket guanaco-reward-models
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: Bucket 's3://guanaco-reward-models/' created
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: uploading /tmp/reward_cache to s3://guanaco-reward-models/dreamgen-opus-v1-2-llama-3-8b-v4_reward
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: cp /tmp/reward_cache/special_tokens_map.json s3://guanaco-reward-models/dreamgen-opus-v1-2-llama-3-8b-v4_reward/special_tokens_map.json
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: cp /tmp/reward_cache/tokenizer_config.json s3://guanaco-reward-models/dreamgen-opus-v1-2-llama-3-8b-v4_reward/tokenizer_config.json
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: cp /tmp/reward_cache/config.json s3://guanaco-reward-models/dreamgen-opus-v1-2-llama-3-8b-v4_reward/config.json
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: cp /tmp/reward_cache/merges.txt s3://guanaco-reward-models/dreamgen-opus-v1-2-llama-3-8b-v4_reward/merges.txt
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: cp /tmp/reward_cache/vocab.json s3://guanaco-reward-models/dreamgen-opus-v1-2-llama-3-8b-v4_reward/vocab.json
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: cp /tmp/reward_cache/tokenizer.json s3://guanaco-reward-models/dreamgen-opus-v1-2-llama-3-8b-v4_reward/tokenizer.json
dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer: cp /tmp/reward_cache/reward.tensors s3://guanaco-reward-models/dreamgen-opus-v1-2-llama-3-8b-v4_reward/reward.tensors
Job dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer completed after 74.08s with status: succeeded
Stopping job with name dreamgen-opus-v1-2-llama-3-8b-v4-mkmlizer
Pipeline stage MKMLizer completed in 79.27s
Running pipeline stage MKMLKubeTemplater
Pipeline stage MKMLKubeTemplater completed in 0.10s
Running pipeline stage ISVCDeployer
Creating inference service dreamgen-opus-v1-2-llama-3-8b-v4
Waiting for inference service dreamgen-opus-v1-2-llama-3-8b-v4 to be ready
Inference service dreamgen-opus-v1-2-llama-3-8b-v4 ready after 30.337286233901978s
Pipeline stage ISVCDeployer completed in 38.16s
Running pipeline stage StressChecker
Received healthy response to inference request in 1.9808545112609863s
Received healthy response to inference request in 1.1877427101135254s
Received healthy response to inference request in 1.133469581604004s
Received healthy response to inference request in 1.1182727813720703s
Received healthy response to inference request in 0.904782772064209s
5 requests
0 failed requests
5th percentile: 0.9474807739257812
10th percentile: 0.9901787757873535
20th percentile: 1.075574779510498
30th percentile: 1.121312141418457
40th percentile: 1.1273908615112305
50th percentile: 1.133469581604004
60th percentile: 1.1551788330078125
70th percentile: 1.1768880844116212
80th percentile: 1.3463650703430177
90th percentile: 1.663609790802002
95th percentile: 1.822232151031494
99th percentile: 1.9491300392150879
mean time: 1.265024471282959
Pipeline stage StressChecker completed in 6.98s
Running pipeline stage DaemonicModelEvalScorer
Pipeline stage DaemonicModelEvalScorer completed in 0.03s
Running pipeline stage DaemonicSafetyScorer
Pipeline stage DaemonicSafetyScorer completed in 0.03s
Running M-Eval for topic stay_in_character
dreamgen-opus-v1-2-llama-3-8b_v4 status is now deployed due to DeploymentManager action
M-Eval Dataset for topic stay_in_character is loaded
dreamgen-opus-v1-2-llama-3-8b_v4 status is now inactive due to auto deactivation removed underperforming models
admin requested tearing down of dreamgen-opus-v1-2-llama-3-8b_v4
Running pipeline stage ISVCDeleter
Checking if service dreamgen-opus-v1-2-llama-3-8b-v4 is running
Tearing down inference service dreamgen-opus-v1-2-llama-3-8b-v4
Toredown service dreamgen-opus-v1-2-llama-3-8b-v4
Pipeline stage ISVCDeleter completed in 4.94s
Running pipeline stage MKMLModelDeleter
Cleaning model data from S3
Cleaning model data from model cache
Deleting key dreamgen-opus-v1-2-llama-3-8b-v4/config.json from bucket guanaco-mkml-models
Deleting key dreamgen-opus-v1-2-llama-3-8b-v4/flywheel_model.0.safetensors from bucket guanaco-mkml-models
Deleting key dreamgen-opus-v1-2-llama-3-8b-v4/special_tokens_map.json from bucket guanaco-mkml-models
Deleting key dreamgen-opus-v1-2-llama-3-8b-v4/tokenizer.json from bucket guanaco-mkml-models
Deleting key dreamgen-opus-v1-2-llama-3-8b-v4/tokenizer_config.json from bucket guanaco-mkml-models
Cleaning model data from model cache
Deleting key dreamgen-opus-v1-2-llama-3-8b-v4_reward/config.json from bucket guanaco-reward-models
Deleting key dreamgen-opus-v1-2-llama-3-8b-v4_reward/merges.txt from bucket guanaco-reward-models
Deleting key dreamgen-opus-v1-2-llama-3-8b-v4_reward/reward.tensors from bucket guanaco-reward-models
Deleting key dreamgen-opus-v1-2-llama-3-8b-v4_reward/special_tokens_map.json from bucket guanaco-reward-models
Deleting key dreamgen-opus-v1-2-llama-3-8b-v4_reward/tokenizer.json from bucket guanaco-reward-models
Deleting key dreamgen-opus-v1-2-llama-3-8b-v4_reward/tokenizer_config.json from bucket guanaco-reward-models
Deleting key dreamgen-opus-v1-2-llama-3-8b-v4_reward/vocab.json from bucket guanaco-reward-models
Pipeline stage MKMLModelDeleter completed in 2.23s
dreamgen-opus-v1-2-llama-3-8b_v4 status is now torndown due to DeploymentManager action

Usage Metrics

Latency Metrics