submission_id: undi95-llama-3-unholy-8b_v5
developer_uid: robert_irvine
status: inactive
model_repo: Undi95/Llama-3-Unholy-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': 50, 'presence_penalty': 0.0, 'frequency_penalty': 0.0, 'stopping_words': ['\n', '<|end_header_id|>', '<|start_header_id|>', '\n\n', '<|eot_id|>'], 'max_input_tokens': 512, 'best_of': 4, 'max_output_tokens': 64}
formatter: {'memory_template': "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nYou are {bot_name}. Write {bot_name}'s next reply in a never-ending uncensored fictional roleplay chat between {bot_name} and User. Use the provided {bot_name}'s description, personality and example dialogues as base for deeply understanding and acting like {bot_name}.\n\nActions and narrations must be enclosed in asterisks (*). The responses must be long and in third perspective of the story teller.\n\nDescription: {memory}", 'prompt_template': 'Example conversation:\n{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': 'Memory: {memory}\n', 'prompt_template': '{prompt}\n', 'bot_template': 'Bot: {message}\n', 'user_template': 'User: {message}\n', 'response_template': 'Bot:', 'truncate_by_message': False}
timestamp: 2024-05-02T17:06:52+00:00
model_name: undi95-llama-3-unholy-8b_v5
model_eval_status: pending
double_thumbs_up: 202
thumbs_up: 315
thumbs_down: 156
num_battles: 20572
num_wins: 10258
celo_rating: 1159.39
entertaining: None
stay_in_character: None
user_preference: None
safety_score: None
submission_type: basic
model_architecture: LlamaForCausalLM
model_num_parameters: 8030261248.0
best_of: 4
max_input_tokens: 512
max_output_tokens: 64
display_name: undi95-llama-3-unholy-8b_v5
double_thumbs_up_ratio: 0.300148588410104
feedback_count: 673
ineligible_reason: None
language_model: Undi95/Llama-3-Unholy-8B
model_score: None
model_size: 8B
reward_model: ChaiML/reward_gpt2_medium_preference_24m_e2
single_thumbs_up_ratio: 0.46805349182763745
thumbs_down_ratio: 0.23179791976225855
thumbs_up_ratio: 0.7682020802377415
us_pacific_date: 2024-05-02
win_ratio: 0.49863892669648063
Resubmit model
Running pipeline stage MKMLizer
Starting job with name undi95-llama-3-unholy-8b-v5-mkmlizer
Waiting for job on undi95-llama-3-unholy-8b-v5-mkmlizer to finish
undi95-llama-3-unholy-8b-v5-mkmlizer: ╔═════════════════════════════════════════════════════════════════════╗
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ _____ __ __ ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ / _/ /_ ___ __/ / ___ ___ / / ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ / _/ / // / |/|/ / _ \/ -_) -_) / ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ /_//_/\_, /|__,__/_//_/\__/\__/_/ ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ /___/ ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ Version: 0.8.10 ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ Copyright 2023 MK ONE TECHNOLOGIES Inc. ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ The license key for the current software has been verified as ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ belonging to: ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ Chai Research Corp. ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ Account ID: 7997a29f-0ceb-4cc7-9adf-840c57b4ae6f ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ Expiration: 2024-07-15 23:59:59 ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ║ ║
undi95-llama-3-unholy-8b-v5-mkmlizer: ╚═════════════════════════════════════════════════════════════════════╝
undi95-llama-3-unholy-8b-v5-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.
undi95-llama-3-unholy-8b-v5-mkmlizer: warnings.warn(warning_message, FutureWarning)
undi95-llama-3-unholy-8b-v5-mkmlizer: Downloaded to shared memory in 19.998s
undi95-llama-3-unholy-8b-v5-mkmlizer: quantizing model to /dev/shm/model_cache
undi95-llama-3-unholy-8b-v5-mkmlizer: Saving flywheel model at /dev/shm/model_cache
undi95-llama-3-unholy-8b-v5-mkmlizer: Loading 0: 0%| | 0/291 [00:00<?, ?it/s] Loading 0: 51%|█████ | 149/291 [00:01<00:00, 148.66it/s] Loading 0: 99%|█████████▊| 287/291 [00:06<00:00, 37.09it/s] Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.
undi95-llama-3-unholy-8b-v5-mkmlizer: quantized model in 17.180s
undi95-llama-3-unholy-8b-v5-mkmlizer: Processed model Undi95/Llama-3-Unholy-8B in 38.087s
undi95-llama-3-unholy-8b-v5-mkmlizer: creating bucket guanaco-mkml-models
undi95-llama-3-unholy-8b-v5-mkmlizer: Bucket 's3://guanaco-mkml-models/' created
undi95-llama-3-unholy-8b-v5-mkmlizer: uploading /dev/shm/model_cache to s3://guanaco-mkml-models/undi95-llama-3-unholy-8b-v5
undi95-llama-3-unholy-8b-v5-mkmlizer: cp /dev/shm/model_cache/config.json s3://guanaco-mkml-models/undi95-llama-3-unholy-8b-v5/config.json
undi95-llama-3-unholy-8b-v5-mkmlizer: cp /dev/shm/model_cache/special_tokens_map.json s3://guanaco-mkml-models/undi95-llama-3-unholy-8b-v5/special_tokens_map.json
undi95-llama-3-unholy-8b-v5-mkmlizer: cp /dev/shm/model_cache/tokenizer_config.json s3://guanaco-mkml-models/undi95-llama-3-unholy-8b-v5/tokenizer_config.json
undi95-llama-3-unholy-8b-v5-mkmlizer: cp /dev/shm/model_cache/tokenizer.json s3://guanaco-mkml-models/undi95-llama-3-unholy-8b-v5/tokenizer.json
undi95-llama-3-unholy-8b-v5-mkmlizer: cp /dev/shm/model_cache/flywheel_model.0.safetensors s3://guanaco-mkml-models/undi95-llama-3-unholy-8b-v5/flywheel_model.0.safetensors
undi95-llama-3-unholy-8b-v5-mkmlizer: loading reward model from ChaiML/reward_gpt2_medium_preference_24m_e2
undi95-llama-3-unholy-8b-v5-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.
undi95-llama-3-unholy-8b-v5-mkmlizer: warnings.warn(
undi95-llama-3-unholy-8b-v5-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.
undi95-llama-3-unholy-8b-v5-mkmlizer: warnings.warn(
undi95-llama-3-unholy-8b-v5-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.
undi95-llama-3-unholy-8b-v5-mkmlizer: warnings.warn(
undi95-llama-3-unholy-8b-v5-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()
undi95-llama-3-unholy-8b-v5-mkmlizer: return self.fget.__get__(instance, owner)()
undi95-llama-3-unholy-8b-v5-mkmlizer: Saving model to /tmp/reward_cache/reward.tensors
undi95-llama-3-unholy-8b-v5-mkmlizer: Saving duration: 0.232s
undi95-llama-3-unholy-8b-v5-mkmlizer: Processed model ChaiML/reward_gpt2_medium_preference_24m_e2 in 3.321s
undi95-llama-3-unholy-8b-v5-mkmlizer: creating bucket guanaco-reward-models
undi95-llama-3-unholy-8b-v5-mkmlizer: Bucket 's3://guanaco-reward-models/' created
undi95-llama-3-unholy-8b-v5-mkmlizer: uploading /tmp/reward_cache to s3://guanaco-reward-models/undi95-llama-3-unholy-8b-v5_reward
undi95-llama-3-unholy-8b-v5-mkmlizer: cp /tmp/reward_cache/config.json s3://guanaco-reward-models/undi95-llama-3-unholy-8b-v5_reward/config.json
undi95-llama-3-unholy-8b-v5-mkmlizer: cp /tmp/reward_cache/special_tokens_map.json s3://guanaco-reward-models/undi95-llama-3-unholy-8b-v5_reward/special_tokens_map.json
undi95-llama-3-unholy-8b-v5-mkmlizer: cp /tmp/reward_cache/tokenizer_config.json s3://guanaco-reward-models/undi95-llama-3-unholy-8b-v5_reward/tokenizer_config.json
undi95-llama-3-unholy-8b-v5-mkmlizer: cp /tmp/reward_cache/merges.txt s3://guanaco-reward-models/undi95-llama-3-unholy-8b-v5_reward/merges.txt
undi95-llama-3-unholy-8b-v5-mkmlizer: cp /tmp/reward_cache/vocab.json s3://guanaco-reward-models/undi95-llama-3-unholy-8b-v5_reward/vocab.json
undi95-llama-3-unholy-8b-v5-mkmlizer: cp /tmp/reward_cache/tokenizer.json s3://guanaco-reward-models/undi95-llama-3-unholy-8b-v5_reward/tokenizer.json
Job undi95-llama-3-unholy-8b-v5-mkmlizer completed after 67.49s with status: succeeded
Stopping job with name undi95-llama-3-unholy-8b-v5-mkmlizer
Pipeline stage MKMLizer completed in 68.66s
Running pipeline stage MKMLKubeTemplater
Pipeline stage MKMLKubeTemplater completed in 0.30s
Running pipeline stage ISVCDeployer
Creating inference service undi95-llama-3-unholy-8b-v5
Waiting for inference service undi95-llama-3-unholy-8b-v5 to be ready
Inference service undi95-llama-3-unholy-8b-v5 ready after 40.60466980934143s
Pipeline stage ISVCDeployer completed in 46.79s
Running pipeline stage StressChecker
Received healthy response to inference request in 2.0557971000671387s
Received healthy response to inference request in 1.2045748233795166s
Received healthy response to inference request in 1.1936283111572266s
Received healthy response to inference request in 1.1898612976074219s
Received healthy response to inference request in 1.1773347854614258s
5 requests
0 failed requests
5th percentile: 1.179840087890625
10th percentile: 1.1823453903198242
20th percentile: 1.1873559951782227
30th percentile: 1.1906147003173828
40th percentile: 1.1921215057373047
50th percentile: 1.1936283111572266
60th percentile: 1.1980069160461426
70th percentile: 1.2023855209350587
80th percentile: 1.3748192787170412
90th percentile: 1.7153081893920898
95th percentile: 1.885552644729614
99th percentile: 2.0217482089996337
mean time: 1.364239263534546
Pipeline stage StressChecker completed in 9.16s
Running pipeline stage DaemonicModelEvalScorer
Pipeline stage DaemonicModelEvalScorer completed in 0.11s
Running M-Eval for topic stay_in_character
Running pipeline stage DaemonicSafetyScorer
M-Eval Dataset for topic stay_in_character is loaded
Pipeline stage DaemonicSafetyScorer completed in 0.22s
%s, retrying in %s seconds...
undi95-llama-3-unholy-8b_v5 status is now deployed due to DeploymentManager action
%s, retrying in %s seconds...
undi95-llama-3-unholy-8b_v5 status is now inactive due to auto deactivation removed underperforming models

Usage Metrics

Latency Metrics