submission_id: jellywibble-qlora-60k-pr_3859_v2
developer_uid: Jellywibble
status: inactive
model_repo: Jellywibble/qlora_60k_pref_data_ep1
reward_repo: ChaiML/reward_gpt2_medium_preference_24m_e2
generation_params: {'temperature': 0.95, 'top_p': 1.0, 'min_p': 0.08, 'top_k': 40, 'presence_penalty': 0.0, 'frequency_penalty': 0.0, 'stopping_words': ['\n', '<|eot_id|>'], 'max_input_tokens': 512, 'best_of': 16, 'max_output_tokens': 64}
formatter: {'memory_template': "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\n{bot_name}'s Persona: {memory}\n\n", 'prompt_template': '{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': 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: 2024-07-03T22:37:22+00:00
model_name: nitral-ai-hathor-l3-8b-v-01_v1
model_group: Jellywibble/qlora_60k_pr
num_battles: 13698
num_wins: 6965
celo_rating: 1194.93
propriety_score: 0.7224127182044888
propriety_total_count: 6416.0
submission_type: basic
model_architecture: LlamaForCausalLM
model_num_parameters: 8030261248.0
best_of: 16
max_input_tokens: 512
max_output_tokens: 64
display_name: nitral-ai-hathor-l3-8b-v-01_v1
ineligible_reason: None
language_model: Jellywibble/qlora_60k_pref_data_ep1
model_size: 8B
reward_model: ChaiML/reward_gpt2_medium_preference_24m_e2
us_pacific_date: 2024-07-03
win_ratio: 0.5084683895459191
Resubmit model
Running pipeline stage MKMLizer
Starting job with name jellywibble-qlora-60k-pr-3859-v2-mkmlizer
Waiting for job on jellywibble-qlora-60k-pr-3859-v2-mkmlizer to finish
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ╔═════════════════════════════════════════════════════════════════════╗
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ _____ __ __ ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ / _/ /_ ___ __/ / ___ ___ / / ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ / _/ / // / |/|/ / _ \/ -_) -_) / ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ /_//_/\_, /|__,__/_//_/\__/\__/_/ ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ /___/ ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ Version: 0.8.14 ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ Copyright 2023 MK ONE TECHNOLOGIES Inc. ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ https://mk1.ai ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ The license key for the current software has been verified as ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ belonging to: ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ Chai Research Corp. ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ Account ID: 7997a29f-0ceb-4cc7-9adf-840c57b4ae6f ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ Expiration: 2024-07-15 23:59:59 ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ║ ║
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: ╚═════════════════════════════════════════════════════════════════════╝
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: Downloaded to shared memory in 53.676s
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: quantizing model to /dev/shm/model_cache
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: Saving flywheel model at /dev/shm/model_cache
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: Loading 0: 0%| | 0/291 [00:00<?, ?it/s] Loading 0: 3%|▎ | 9/291 [00:00<00:03, 87.36it/s] Loading 0: 7%|▋ | 21/291 [00:00<00:02, 97.02it/s] Loading 0: 11%|█ | 31/291 [00:00<00:02, 89.69it/s] Loading 0: 15%|█▌ | 44/291 [00:00<00:02, 102.50it/s] Loading 0: 20%|█▉ | 57/291 [00:00<00:02, 111.48it/s] Loading 0: 24%|██▎ | 69/291 [00:00<00:02, 101.07it/s] Loading 0: 27%|██▋ | 80/291 [00:00<00:02, 100.59it/s] Loading 0: 31%|███▏ | 91/291 [00:01<00:03, 56.49it/s] Loading 0: 35%|███▌ | 102/291 [00:01<00:02, 64.02it/s] Loading 0: 38%|███▊ | 112/291 [00:01<00:02, 69.34it/s] Loading 0: 42%|████▏ | 122/291 [00:01<00:02, 75.97it/s] Loading 0: 47%|████▋ | 138/291 [00:01<00:01, 92.27it/s] Loading 0: 51%|█████ | 149/291 [00:01<00:01, 91.34it/s] Loading 0: 55%|█████▌ | 161/291 [00:01<00:01, 96.76it/s] Loading 0: 59%|█████▉ | 172/291 [00:01<00:01, 97.42it/s] Loading 0: 63%|██████▎ | 183/291 [00:02<00:01, 96.93it/s] Loading 0: 67%|██████▋ | 194/291 [00:02<00:01, 56.08it/s] Loading 0: 71%|███████ | 206/291 [00:02<00:01, 67.15it/s] Loading 0: 75%|███████▌ | 219/291 [00:02<00:00, 75.83it/s] Loading 0: 79%|███████▊ | 229/291 [00:02<00:00, 77.20it/s] Loading 0: 82%|████████▏ | 239/291 [00:02<00:00, 76.35it/s] Loading 0: 87%|████████▋ | 254/291 [00:03<00:00, 93.26it/s] Loading 0: 91%|█████████ | 265/291 [00:03<00:00, 88.93it/s] Loading 0: 95%|█████████▍| 275/291 [00:03<00:00, 88.94it/s] Loading 0: 99%|█████████▊| 287/291 [00:09<00:00, 5.95it/s] Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: quantized model in 25.878s
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: Processed model Jellywibble/qlora_60k_pref_data_ep1 in 79.554s
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: creating bucket guanaco-mkml-models
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: Bucket 's3://guanaco-mkml-models/' created
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: uploading /dev/shm/model_cache to s3://guanaco-mkml-models/jellywibble-qlora-60k-pr-3859-v2
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: cp /dev/shm/model_cache/config.json s3://guanaco-mkml-models/jellywibble-qlora-60k-pr-3859-v2/config.json
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: cp /dev/shm/model_cache/special_tokens_map.json s3://guanaco-mkml-models/jellywibble-qlora-60k-pr-3859-v2/special_tokens_map.json
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: cp /dev/shm/model_cache/tokenizer_config.json s3://guanaco-mkml-models/jellywibble-qlora-60k-pr-3859-v2/tokenizer_config.json
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: cp /dev/shm/model_cache/tokenizer.json s3://guanaco-mkml-models/jellywibble-qlora-60k-pr-3859-v2/tokenizer.json
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: cp /dev/shm/model_cache/flywheel_model.0.safetensors s3://guanaco-mkml-models/jellywibble-qlora-60k-pr-3859-v2/flywheel_model.0.safetensors
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: loading reward model from ChaiML/reward_gpt2_medium_preference_24m_e2
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: /opt/conda/lib/python3.10/site-packages/transformers/models/auto/configuration_auto.py:919: FutureWarning: The `use_auth_token` argument is deprecated and will be removed in v5 of Transformers. Please use `token` instead.
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: warnings.warn(
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: /opt/conda/lib/python3.10/site-packages/huggingface_hub/file_download.py:1132: FutureWarning: `resume_download` is deprecated and will be removed in version 1.0.0. Downloads always resume when possible. If you want to force a new download, use `force_download=True`.
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: warnings.warn(
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: /opt/conda/lib/python3.10/site-packages/transformers/models/auto/tokenization_auto.py:769: FutureWarning: The `use_auth_token` argument is deprecated and will be removed in v5 of Transformers. Please use `token` instead.
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: warnings.warn(
jellywibble-qlora-60k-pr-3859-v2-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.
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: warnings.warn(
jellywibble-qlora-60k-pr-3859-v2-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()
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: return self.fget.__get__(instance, owner)()
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: Saving model to /tmp/reward_cache/reward.tensors
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: Saving duration: 0.420s
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: Processed model ChaiML/reward_gpt2_medium_preference_24m_e2 in 12.402s
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: creating bucket guanaco-reward-models
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: Bucket 's3://guanaco-reward-models/' created
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: uploading /tmp/reward_cache to s3://guanaco-reward-models/jellywibble-qlora-60k-pr-3859-v2_reward
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: cp /tmp/reward_cache/config.json s3://guanaco-reward-models/jellywibble-qlora-60k-pr-3859-v2_reward/config.json
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: cp /tmp/reward_cache/special_tokens_map.json s3://guanaco-reward-models/jellywibble-qlora-60k-pr-3859-v2_reward/special_tokens_map.json
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: cp /tmp/reward_cache/tokenizer_config.json s3://guanaco-reward-models/jellywibble-qlora-60k-pr-3859-v2_reward/tokenizer_config.json
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: cp /tmp/reward_cache/merges.txt s3://guanaco-reward-models/jellywibble-qlora-60k-pr-3859-v2_reward/merges.txt
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: cp /tmp/reward_cache/vocab.json s3://guanaco-reward-models/jellywibble-qlora-60k-pr-3859-v2_reward/vocab.json
jellywibble-qlora-60k-pr-3859-v2-mkmlizer: cp /tmp/reward_cache/tokenizer.json s3://guanaco-reward-models/jellywibble-qlora-60k-pr-3859-v2_reward/tokenizer.json
Retrying (%r) after connection broken by '%r': %s
Job jellywibble-qlora-60k-pr-3859-v2-mkmlizer completed after 124.74s with status: succeeded
Stopping job with name jellywibble-qlora-60k-pr-3859-v2-mkmlizer
Pipeline stage MKMLizer completed in 125.77s
Running pipeline stage MKMLKubeTemplater
Pipeline stage MKMLKubeTemplater completed in 0.14s
Running pipeline stage ISVCDeployer
Creating inference service jellywibble-qlora-60k-pr-3859-v2
Waiting for inference service jellywibble-qlora-60k-pr-3859-v2 to be ready
Inference service jellywibble-qlora-60k-pr-3859-v2 ready after 50.2473361492157s
Pipeline stage ISVCDeployer completed in 56.95s
Running pipeline stage StressChecker
Received healthy response to inference request in 2.090834856033325s
Received healthy response to inference request in 1.3392767906188965s
Received healthy response to inference request in 1.3229241371154785s
Received healthy response to inference request in 1.2923369407653809s
Received healthy response to inference request in 1.362818717956543s
5 requests
0 failed requests
5th percentile: 1.2984543800354005
10th percentile: 1.3045718193054199
20th percentile: 1.316806697845459
30th percentile: 1.326194667816162
40th percentile: 1.3327357292175293
50th percentile: 1.3392767906188965
60th percentile: 1.348693561553955
70th percentile: 1.3581103324890136
80th percentile: 1.5084219455718995
90th percentile: 1.7996284008026124
95th percentile: 1.9452316284179687
99th percentile: 2.061714210510254
mean time: 1.481638288497925
Pipeline stage StressChecker completed in 8.17s
jellywibble-qlora-60k-pr_3859_v2 status is now deployed due to DeploymentManager action
jellywibble-qlora-60k-pr_3859_v2 status is now inactive due to auto deactivation removed underperforming models

Usage Metrics

Latency Metrics