rica40325-feedback-dpo-6

submission_id: rica40325-feedback-dpo-6_v1

developer_uid: rica40325

alignment_samples: 11426

alignment_score: -0.052296476470748414

best_of: 16

celo_rating: 875.75

display_name: rica40325-feedback-dpo-5_v1

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}

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': 16, 'max_output_tokens': 64}

gpu_counts: {'NVIDIA RTX A5000': 1}

is_internal_developer: False

language_model: rica40325/feedback_dpo_6

latencies: [{'batch_size': 1, 'throughput': 0.9149594286351588, 'latency_mean': 1.092846233844757, 'latency_p50': 1.0910205841064453, 'latency_p90': 1.2265576601028443}, {'batch_size': 4, 'throughput': 1.8494913278647844, 'latency_mean': 2.151351844072342, 'latency_p50': 2.153593897819519, 'latency_p90': 2.404458522796631}, {'batch_size': 5, 'throughput': 1.8984367282608081, 'latency_mean': 2.6197708296775817, 'latency_p50': 2.6407910585403442, 'latency_p90': 2.9384270191192625}, {'batch_size': 8, 'throughput': 2.0218104272950037, 'latency_mean': 3.9289731550216676, 'latency_p50': 3.909578323364258, 'latency_p90': 4.484247875213623}, {'batch_size': 10, 'throughput': 2.034897569776057, 'latency_mean': 4.863874319791794, 'latency_p50': 4.863814830780029, 'latency_p90': 5.5642486095428465}, {'batch_size': 12, 'throughput': 2.0552831386624555, 'latency_mean': 5.75637845993042, 'latency_p50': 5.834072947502136, 'latency_p90': 6.562680912017822}, {'batch_size': 15, 'throughput': 2.041589135973876, 'latency_mean': 7.206974573135376, 'latency_p50': 7.294035911560059, 'latency_p90': 8.009525489807128}]

max_input_tokens: 512

max_output_tokens: 64

model_architecture: LlamaForCausalLM

model_group: rica40325/feedback_dpo_6

model_name: rica40325-feedback-dpo-5_v1

model_num_parameters: 8030261248.0

model_repo: rica40325/feedback_dpo_6

model_size: 8B

num_battles: 11426

num_wins: 1517

propriety_score: 0.6158008658008658

propriety_total_count: 924.0

ranking_group: single

status: inactive

submission_type: basic

throughput_3p7s: 2.02

timestamp: 2024-09-10T09:02:07+00:00

us_pacific_date: 2024-09-10

win_ratio: 0.13276737265884825

Download Preference Data

Resubmit model

Shutdown handler not registered because Python interpreter is not running in the main thread
run pipeline %s
run pipeline stage %s
Running pipeline stage MKMLizer
Starting job with name rica40325-feedback-dpo-6-v1-mkmlizer
Waiting for job on rica40325-feedback-dpo-6-v1-mkmlizer to finish
rica40325-feedback-dpo-6-v1-mkmlizer: ╔═════════════════════════════════════════════════════════════════════╗
rica40325-feedback-dpo-6-v1-mkmlizer: ║     _____            __           __                                ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║    / _/ /_ ___    __/ /  ___ ___ / /                                ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║   / _/ / // / |/|/ / _ \/ -_) -_) /                                 ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║  /_//_/\_, /|__,__/_//_/\__/\__/_/                                  ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║       /___/                                                         ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║                                                                     ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║  Version: 0.10.1                                                    ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║  Copyright 2023 MK ONE TECHNOLOGIES Inc.                            ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║  https://mk1.ai                                                     ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║                                                                     ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║  The license key for the current software has been verified as      ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║  belonging to:                                                      ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║                                                                     ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║  Chai Research Corp.                                                ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║  Account ID: 7997a29f-0ceb-4cc7-9adf-840c57b4ae6f                   ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║  Expiration: 2024-10-15 23:59:59                                    ║
rica40325-feedback-dpo-6-v1-mkmlizer: ║                                                                     ║
rica40325-feedback-dpo-6-v1-mkmlizer: ╚═════════════════════════════════════════════════════════════════════╝
rica40325-feedback-dpo-6-v1-mkmlizer: Downloaded to shared memory in 58.975s
rica40325-feedback-dpo-6-v1-mkmlizer: quantizing model to /dev/shm/model_cache, profile:s0, folder:/tmp/tmpr2gzcb1o, device:0
rica40325-feedback-dpo-6-v1-mkmlizer: Saving flywheel model at /dev/shm/model_cache
Connection pool is full, discarding connection: %s. Connection pool size: %s
Connection pool is full, discarding connection: %s. Connection pool size: %s
Connection pool is full, discarding connection: %s. Connection pool size: %s
Connection pool is full, discarding connection: %s. Connection pool size: %s
Connection pool is full, discarding connection: %s. Connection pool size: %s
Connection pool is full, discarding connection: %s. Connection pool size: %s
rica40325-feedback-dpo-6-v1-mkmlizer: quantized model in 28.948s
rica40325-feedback-dpo-6-v1-mkmlizer: Processed model rica40325/feedback_dpo_6 in 87.923s
rica40325-feedback-dpo-6-v1-mkmlizer: creating bucket guanaco-mkml-models
rica40325-feedback-dpo-6-v1-mkmlizer: Bucket 's3://guanaco-mkml-models/' created
rica40325-feedback-dpo-6-v1-mkmlizer: uploading /dev/shm/model_cache to s3://guanaco-mkml-models/rica40325-feedback-dpo-6-v1
rica40325-feedback-dpo-6-v1-mkmlizer: cp /dev/shm/model_cache/config.json s3://guanaco-mkml-models/rica40325-feedback-dpo-6-v1/config.json
rica40325-feedback-dpo-6-v1-mkmlizer: cp /dev/shm/model_cache/special_tokens_map.json s3://guanaco-mkml-models/rica40325-feedback-dpo-6-v1/special_tokens_map.json
rica40325-feedback-dpo-6-v1-mkmlizer: cp /dev/shm/model_cache/tokenizer_config.json s3://guanaco-mkml-models/rica40325-feedback-dpo-6-v1/tokenizer_config.json
rica40325-feedback-dpo-6-v1-mkmlizer: cp /dev/shm/model_cache/tokenizer.json s3://guanaco-mkml-models/rica40325-feedback-dpo-6-v1/tokenizer.json
rica40325-feedback-dpo-6-v1-mkmlizer: cp /dev/shm/model_cache/flywheel_model.0.safetensors s3://guanaco-mkml-models/rica40325-feedback-dpo-6-v1/flywheel_model.0.safetensors
rica40325-feedback-dpo-6-v1-mkmlizer: 
Loading 0:   0%|          | 0/291 [00:00<?, ?it/s]
Loading 0:   2%|▏         | 5/291 [00:00<00:10, 26.22it/s]
Loading 0:   4%|▍         | 12/291 [00:00<00:07, 37.00it/s]
Loading 0:   5%|▌         | 16/291 [00:00<00:07, 34.39it/s]
Loading 0:   7%|▋         | 21/291 [00:00<00:07, 38.08it/s]
Loading 0:   9%|▊         | 25/291 [00:00<00:07, 34.76it/s]
Loading 0:  10%|█         | 30/291 [00:00<00:06, 38.28it/s]
Loading 0:  12%|█▏        | 34/291 [00:01<00:09, 27.08it/s]
Loading 0:  13%|█▎        | 38/291 [00:01<00:09, 27.73it/s]
Loading 0:  14%|█▍        | 42/291 [00:01<00:09, 26.19it/s]
Loading 0:  16%|█▋        | 48/291 [00:01<00:07, 31.94it/s]
Loading 0:  18%|█▊        | 52/291 [00:01<00:07, 30.77it/s]
Loading 0:  20%|█▉        | 57/291 [00:01<00:07, 32.78it/s]
Loading 0:  21%|██        | 61/291 [00:01<00:07, 31.38it/s]
Loading 0:  23%|██▎       | 66/291 [00:02<00:06, 32.90it/s]
Loading 0:  24%|██▍       | 70/291 [00:02<00:06, 31.73it/s]
Loading 0:  25%|██▌       | 74/291 [00:02<00:06, 32.13it/s]
Loading 0:  27%|██▋       | 78/291 [00:02<00:06, 32.02it/s]
Loading 0:  28%|██▊       | 82/291 [00:02<00:08, 23.45it/s]
Loading 0:  29%|██▉       | 85/291 [00:02<00:08, 24.24it/s]
Loading 0:  31%|███       | 90/291 [00:02<00:06, 28.77it/s]
Loading 0:  32%|███▏      | 94/291 [00:03<00:06, 28.37it/s]
Loading 0:  34%|███▍      | 99/291 [00:03<00:05, 32.13it/s]
Loading 0:  35%|███▌      | 103/291 [00:03<00:06, 30.70it/s]
Loading 0:  37%|███▋      | 108/291 [00:03<00:05, 32.61it/s]
Loading 0:  38%|███▊      | 112/291 [00:03<00:05, 31.88it/s]
Loading 0:  40%|███▉      | 116/291 [00:03<00:05, 32.70it/s]
Loading 0:  42%|████▏     | 122/291 [00:03<00:04, 36.98it/s]
Loading 0:  44%|████▎     | 127/291 [00:04<00:04, 34.86it/s]
Loading 0:  46%|████▌     | 133/291 [00:04<00:05, 30.11it/s]
Loading 0:  47%|████▋     | 137/291 [00:04<00:05, 29.95it/s]
Loading 0:  48%|████▊     | 141/291 [00:04<00:05, 28.08it/s]
Loading 0:  51%|█████     | 147/291 [00:04<00:04, 32.91it/s]
Loading 0:  52%|█████▏    | 151/291 [00:04<00:04, 31.78it/s]
Loading 0:  54%|█████▎    | 156/291 [00:04<00:03, 34.56it/s]
Loading 0:  55%|█████▍    | 160/291 [00:05<00:04, 32.65it/s]
Loading 0:  57%|█████▋    | 165/291 [00:05<00:03, 33.68it/s]
Loading 0:  58%|█████▊    | 169/291 [00:05<00:03, 32.47it/s]
Loading 0:  60%|█████▉    | 174/291 [00:05<00:03, 34.68it/s]
Loading 0:  61%|██████    | 178/291 [00:05<00:03, 32.96it/s]
Loading 0:  63%|██████▎   | 184/291 [00:05<00:02, 38.98it/s]
Loading 0:  65%|██████▍   | 189/291 [00:06<00:04, 24.10it/s]
Loading 0:  67%|██████▋   | 194/291 [00:06<00:03, 25.40it/s]
Loading 0:  69%|██████▉   | 201/291 [00:06<00:02, 31.98it/s]
Loading 0:  70%|███████   | 205/291 [00:06<00:02, 31.17it/s]
Loading 0:  72%|███████▏  | 210/291 [00:06<00:02, 33.87it/s]
Loading 0:  74%|███████▎  | 214/291 [00:06<00:02, 32.73it/s]
Loading 0:  75%|███████▌  | 219/291 [00:06<00:02, 33.66it/s]
Loading 0:  77%|███████▋  | 223/291 [00:07<00:02, 32.36it/s]
Loading 0:  78%|███████▊  | 227/291 [00:07<00:02, 31.34it/s]
Loading 0:  79%|███████▉  | 231/291 [00:07<00:01, 31.94it/s]
Loading 0:  81%|████████  | 235/291 [00:07<00:02, 23.98it/s]
Loading 0:  82%|████████▏ | 239/291 [00:07<00:02, 23.97it/s]
Loading 0:  85%|████████▍ | 246/291 [00:07<00:01, 32.12it/s]
Loading 0:  86%|████████▌ | 250/291 [00:08<00:01, 31.94it/s]
Loading 0:  88%|████████▊ | 255/291 [00:08<00:01, 34.43it/s]
Loading 0:  89%|████████▉ | 259/291 [00:08<00:00, 33.46it/s]
Loading 0:  91%|█████████ | 264/291 [00:08<00:00, 36.63it/s]
Loading 0:  92%|█████████▏| 268/291 [00:08<00:00, 35.00it/s]
Loading 0:  94%|█████████▍| 273/291 [00:08<00:00, 37.61it/s]
Loading 0:  95%|█████████▌| 277/291 [00:08<00:00, 34.95it/s]
Loading 0:  97%|█████████▋| 281/291 [00:08<00:00, 34.68it/s]
Loading 0:  98%|█████████▊| 286/291 [00:14<00:01,  2.58it/s]
Loading 0:  99%|█████████▉| 289/291 [00:14<00:00,  3.21it/s]
                                                            
Job rica40325-feedback-dpo-6-v1-mkmlizer completed after 114.76s with status: succeeded
Stopping job with name rica40325-feedback-dpo-6-v1-mkmlizer
Pipeline stage MKMLizer completed in 115.99s
run pipeline stage %s
Running pipeline stage MKMLTemplater
Pipeline stage MKMLTemplater completed in 0.19s
run pipeline stage %s
Running pipeline stage MKMLDeployer
Creating inference service rica40325-feedback-dpo-6-v1
Waiting for inference service rica40325-feedback-dpo-6-v1 to be ready
Failed to get response for submission blend_hokok_2024-09-09: ('http://chaiml-llama-8b-pairwis-8189-v19-predictor.tenant-chaiml-guanaco.k.chaiverse.com/v1/models/GPT-J-6B-lit-v2:predict', 'read tcp 127.0.0.1:54784->127.0.0.1:8080: read: connection reset by peer\n')
Connection pool is full, discarding connection: %s. Connection pool size: %s
Connection pool is full, discarding connection: %s. Connection pool size: %s
Failed to get response for submission blend_puheb_2024-09-09: ('http://zonemercy-lexical-nemo-1518-v18-predictor.tenant-chaiml-guanaco.k.chaiverse.com/v1/models/GPT-J-6B-lit-v2:predict', 'read tcp 127.0.0.1:41566->127.0.0.1:8080: read: connection reset by peer\n')
Inference service rica40325-feedback-dpo-6-v1 ready after 151.51099371910095s
Pipeline stage MKMLDeployer completed in 152.07s
run pipeline stage %s
Running pipeline stage StressChecker
Received healthy response to inference request in 2.0474307537078857s
Received healthy response to inference request in 1.6481406688690186s
Received healthy response to inference request in 1.352252721786499s
Received healthy response to inference request in 1.6556634902954102s
Received healthy response to inference request in 2.2182955741882324s
5 requests
0 failed requests
5th percentile: 1.4114303112030029
10th percentile: 1.4706079006195067
20th percentile: 1.5889630794525147
30th percentile: 1.6496452331542968
40th percentile: 1.6526543617248535
50th percentile: 1.6556634902954102
60th percentile: 1.8123703956604005
70th percentile: 1.9690773010253906
80th percentile: 2.0816037178039553
90th percentile: 2.149949645996094
95th percentile: 2.184122610092163
99th percentile: 2.2114609813690187
mean time: 1.7843566417694092
Pipeline stage StressChecker completed in 9.91s
run pipeline stage %s
Running pipeline stage TriggerMKMLProfilingPipeline
run_pipeline:run_in_cloud %s
starting trigger_guanaco_pipeline args=%s
Pipeline stage TriggerMKMLProfilingPipeline completed in 5.32s
Shutdown handler de-registered
rica40325-feedback-dpo-6_v1 status is now deployed due to DeploymentManager action
Shutdown handler registered
run pipeline %s
run pipeline stage %s
Running pipeline stage MKMLProfilerDeleter
Skipping teardown as no inference service was successfully deployed
Pipeline stage MKMLProfilerDeleter completed in 0.14s
run pipeline stage %s
Running pipeline stage MKMLProfilerTemplater
Pipeline stage MKMLProfilerTemplater completed in 0.10s
run pipeline stage %s
Running pipeline stage MKMLProfilerDeployer
Creating inference service rica40325-feedback-dpo-6-v1-profiler
Waiting for inference service rica40325-feedback-dpo-6-v1-profiler to be ready
Inference service rica40325-feedback-dpo-6-v1-profiler ready after 150.358904838562s
Pipeline stage MKMLProfilerDeployer completed in 150.70s
run pipeline stage %s
Running pipeline stage MKMLProfilerRunner
kubectl cp /code/guanaco/guanaco_inference_services/src/inference_scripts tenant-chaiml-guanaco/rica40325-feedback-dpo-6-v1-profiler-predictor-00001-deplolhw6n:/code/chaiverse_profiler_1725959400 --namespace tenant-chaiml-guanaco
kubectl exec -it rica40325-feedback-dpo-6-v1-profiler-predictor-00001-deplolhw6n --namespace tenant-chaiml-guanaco -- sh -c 'cd /code/chaiverse_profiler_1725959400 && python profiles.py profile --best_of_n 16 --auto_batch 5 --batches 1,5,10,15,20,25,30,35,40,45,50,55,60,65,70,75,80,85,90,95,100,105,110,115,120,125,130,135,140,145,150,155,160,165,170,175,180,185,190,195 --samples 200 --input_tokens 512 --output_tokens 64 --summary /code/chaiverse_profiler_1725959400/summary.json'
kubectl exec -it rica40325-feedback-dpo-6-v1-profiler-predictor-00001-deplolhw6n --namespace tenant-chaiml-guanaco -- bash -c 'cat /code/chaiverse_profiler_1725959400/summary.json'
Pipeline stage MKMLProfilerRunner completed in 830.32s
run pipeline stage %s
Running pipeline stage MKMLProfilerDeleter
Checking if service rica40325-feedback-dpo-6-v1-profiler is running
Tearing down inference service rica40325-feedback-dpo-6-v1-profiler
Service rica40325-feedback-dpo-6-v1-profiler has been torndown
Pipeline stage MKMLProfilerDeleter completed in 1.86s
Shutdown handler de-registered
rica40325-feedback-dpo-6_v1 status is now inactive due to auto deactivation removed underperforming models