chaiml-albert-dpo-0912-v_2161

submission_id: chaiml-albert-dpo-0912-v_2161_v1

developer_uid: chai_backend_admin

alignment_samples: 10975

alignment_score: -0.20477587824417373

best_of: 8

celo_rating: 1245.94

display_name: edit_variety_conflict

formatter: {'memory_template': '', 'prompt_template': '', 'bot_template': '{bot_name}: {message}\n', 'user_template': '{user_name}: {message}\n', 'response_template': '{bot_name}:', 'truncate_by_message': False}

generation_params: {'temperature': 0.9, 'top_p': 1.0, 'min_p': 0.05, 'top_k': 80, 'presence_penalty': 0.0, 'frequency_penalty': 0.0, 'stopping_words': ['\n', '</s>', '###', 'Bot:', 'User:', 'You:', '<|im_end|>'], 'max_input_tokens': 1024, 'best_of': 8, 'max_output_tokens': 64}

gpu_counts: {'NVIDIA RTX A5000': 1}

is_internal_developer: True

language_model: ChaiML/albert_dpo_0912_virgo_edit_variety_conflict

latencies: [{'batch_size': 1, 'throughput': 0.6198261361046796, 'latency_mean': 1.613294153213501, 'latency_p50': 1.610878825187683, 'latency_p90': 1.798569130897522}, {'batch_size': 3, 'throughput': 1.0929839501923861, 'latency_mean': 2.7388402807712553, 'latency_p50': 2.7237930297851562, 'latency_p90': 3.011830472946167}, {'batch_size': 5, 'throughput': 1.235698693419649, 'latency_mean': 4.018700484037399, 'latency_p50': 4.042511820793152, 'latency_p90': 4.511973142623901}, {'batch_size': 6, 'throughput': 1.2753544257591805, 'latency_mean': 4.688752344846725, 'latency_p50': 4.691368937492371, 'latency_p90': 5.345764374732971}, {'batch_size': 8, 'throughput': 1.2585300148800336, 'latency_mean': 6.327800551652908, 'latency_p50': 6.327863097190857, 'latency_p90': 7.308955717086792}, {'batch_size': 10, 'throughput': 1.2093472752894567, 'latency_mean': 8.210466510057449, 'latency_p50': 8.281776428222656, 'latency_p90': 9.165217518806458}]

max_input_tokens: 1024

max_output_tokens: 64

model_architecture: MistralForCausalLM

model_group: ChaiML/albert_dpo_0912_v

model_name: edit_variety_conflict

model_num_parameters: 12772070400.0

model_repo: ChaiML/albert_dpo_0912_virgo_edit_variety_conflict

model_size: 13B

num_battles: 10974

num_wins: 5486

propriety_score: 0.7515991471215352

propriety_total_count: 938.0

ranking_group: single

status: inactive

submission_type: basic

throughput_3p7s: 1.22

timestamp: 2024-09-13T19:24:10+00:00

us_pacific_date: 2024-09-13

win_ratio: 0.49990887552396573

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 chaiml-albert-dpo-0912-v-2161-v1-mkmlizer
Waiting for job on chaiml-albert-dpo-0912-v-2161-v1-mkmlizer to finish
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ╔═════════════════════════════════════════════════════════════════════╗
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║     _____            __           __                                ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║    / _/ /_ ___    __/ /  ___ ___ / /                                ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║   / _/ / // / |/|/ / _ \/ -_) -_) /                                 ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║  /_//_/\_, /|__,__/_//_/\__/\__/_/                                  ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║       /___/                                                         ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║                                                                     ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║  Version: 0.10.1                                                    ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║  Copyright 2023 MK ONE TECHNOLOGIES Inc.                            ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║  https://mk1.ai                                                     ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║                                                                     ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║  The license key for the current software has been verified as      ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║  belonging to:                                                      ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║                                                                     ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║  Chai Research Corp.                                                ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║  Account ID: 7997a29f-0ceb-4cc7-9adf-840c57b4ae6f                   ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║  Expiration: 2024-10-15 23:59:59                                    ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ║                                                                     ║
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: ╚═════════════════════════════════════════════════════════════════════╝
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: Downloaded to shared memory in 102.345s
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: quantizing model to /dev/shm/model_cache, profile:s0, folder:/tmp/tmpq7rfwym1, device:0
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: Saving flywheel model at /dev/shm/model_cache
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: quantized model in 40.733s
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: Processed model ChaiML/albert_dpo_0912_virgo_edit_variety_conflict in 143.078s
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: creating bucket guanaco-mkml-models
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: Bucket 's3://guanaco-mkml-models/' created
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: uploading /dev/shm/model_cache to s3://guanaco-mkml-models/chaiml-albert-dpo-0912-v-2161-v1
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: cp /dev/shm/model_cache/config.json s3://guanaco-mkml-models/chaiml-albert-dpo-0912-v-2161-v1/config.json
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: cp /dev/shm/model_cache/special_tokens_map.json s3://guanaco-mkml-models/chaiml-albert-dpo-0912-v-2161-v1/special_tokens_map.json
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: cp /dev/shm/model_cache/tokenizer_config.json s3://guanaco-mkml-models/chaiml-albert-dpo-0912-v-2161-v1/tokenizer_config.json
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: cp /dev/shm/model_cache/tokenizer.json s3://guanaco-mkml-models/chaiml-albert-dpo-0912-v-2161-v1/tokenizer.json
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: cp /dev/shm/model_cache/flywheel_model.0.safetensors s3://guanaco-mkml-models/chaiml-albert-dpo-0912-v-2161-v1/flywheel_model.0.safetensors
chaiml-albert-dpo-0912-v-2161-v1-mkmlizer: 
Loading 0:   0%|          | 0/363 [00:00<?, ?it/s]
Loading 0:   1%|▏         | 5/363 [00:00<00:15, 22.63it/s]
Loading 0:   3%|▎         | 10/363 [00:00<00:12, 28.12it/s]
Loading 0:   4%|▍         | 14/363 [00:00<00:14, 24.47it/s]
Loading 0:   5%|▌         | 19/363 [00:00<00:11, 30.66it/s]
Loading 0:   6%|▋         | 23/363 [00:00<00:16, 20.79it/s]
Loading 0:   7%|▋         | 26/363 [00:01<00:17, 19.13it/s]
Loading 0:   9%|▊         | 31/363 [00:01<00:13, 24.86it/s]
Loading 0:  10%|▉         | 35/363 [00:01<00:12, 26.38it/s]
Loading 0:  11%|█         | 39/363 [00:01<00:11, 27.81it/s]
Loading 0:  12%|█▏        | 43/363 [00:01<00:11, 26.87it/s]
Loading 0:  13%|█▎        | 48/363 [00:01<00:10, 29.86it/s]
Loading 0:  14%|█▍        | 52/363 [00:01<00:11, 28.10it/s]
Loading 0:  15%|█▌        | 55/363 [00:02<00:11, 27.15it/s]
Loading 0:  17%|█▋        | 60/363 [00:02<00:10, 29.85it/s]
Loading 0:  18%|█▊        | 64/363 [00:02<00:14, 20.43it/s]
Loading 0:  20%|█▉        | 71/363 [00:02<00:10, 27.48it/s]
Loading 0:  21%|██        | 75/363 [00:02<00:10, 26.82it/s]
Loading 0:  22%|██▏       | 79/363 [00:03<00:10, 26.12it/s]
Loading 0:  23%|██▎       | 84/363 [00:03<00:09, 28.40it/s]
Loading 0:  24%|██▍       | 88/363 [00:03<00:09, 28.00it/s]
Loading 0:  26%|██▌       | 93/363 [00:03<00:08, 30.95it/s]
Loading 0:  27%|██▋       | 97/363 [00:03<00:08, 30.40it/s]
Loading 0:  28%|██▊       | 101/363 [00:03<00:10, 25.79it/s]
Loading 0:  29%|██▊       | 104/363 [00:04<00:11, 22.35it/s]
Loading 0:  31%|███       | 111/363 [00:04<00:08, 29.98it/s]
Loading 0:  32%|███▏      | 115/363 [00:04<00:08, 28.39it/s]
Loading 0:  33%|███▎      | 120/363 [00:04<00:07, 31.30it/s]
Loading 0:  34%|███▍      | 124/363 [00:04<00:07, 30.48it/s]
Loading 0:  35%|███▌      | 128/363 [00:04<00:07, 31.07it/s]
Loading 0:  36%|███▋      | 132/363 [00:04<00:08, 26.91it/s]
Loading 0:  38%|███▊      | 137/363 [00:05<00:07, 30.16it/s]
Loading 0:  39%|███▉      | 142/363 [00:05<00:08, 25.98it/s]
Loading 0:  40%|███▉      | 145/363 [00:05<00:09, 23.92it/s]
Loading 0:  41%|████      | 149/363 [00:05<00:09, 22.81it/s]
Loading 0:  43%|████▎     | 156/363 [00:05<00:06, 30.03it/s]
Loading 0:  44%|████▍     | 160/363 [00:05<00:07, 28.78it/s]
Loading 0:  45%|████▌     | 165/363 [00:06<00:06, 31.13it/s]
Loading 0:  47%|████▋     | 169/363 [00:06<00:06, 29.83it/s]
Loading 0:  48%|████▊     | 174/363 [00:06<00:06, 31.18it/s]
Loading 0:  49%|████▉     | 178/363 [00:06<00:06, 29.61it/s]
Loading 0:  50%|█████     | 182/363 [00:06<00:07, 24.18it/s]
Loading 0:  51%|█████     | 185/363 [00:06<00:08, 21.47it/s]
Loading 0:  53%|█████▎    | 192/363 [00:07<00:05, 28.59it/s]
Loading 0:  54%|█████▍    | 196/363 [00:07<00:05, 28.07it/s]
Loading 0:  55%|█████▌    | 201/363 [00:07<00:05, 30.55it/s]
Loading 0:  56%|█████▋    | 205/363 [00:07<00:05, 29.02it/s]
Loading 0:  58%|█████▊    | 210/363 [00:07<00:04, 31.81it/s]
Loading 0:  59%|█████▉    | 214/363 [00:07<00:05, 29.25it/s]
Loading 0:  60%|██████    | 218/363 [00:07<00:04, 29.17it/s]
Loading 0:  61%|██████▏   | 223/363 [00:08<00:05, 25.59it/s]
Loading 0:  62%|██████▏   | 226/363 [00:08<00:05, 24.02it/s]
Loading 0:  63%|██████▎   | 230/363 [00:08<00:05, 22.67it/s]
Loading 0:  65%|██████▍   | 235/363 [00:08<00:04, 27.42it/s]
Loading 0:  66%|██████▌   | 239/363 [00:08<00:05, 24.78it/s]
Loading 0:  67%|██████▋   | 244/363 [00:08<00:04, 29.38it/s]
Loading 0:  68%|██████▊   | 248/363 [00:09<00:04, 26.30it/s]
Loading 0:  70%|███████   | 255/363 [00:09<00:03, 32.35it/s]
Loading 0:  71%|███████▏  | 259/363 [00:09<00:03, 30.24it/s]
Loading 0:  72%|███████▏  | 263/363 [00:09<00:04, 24.90it/s]
Loading 0:  73%|███████▎  | 266/363 [00:09<00:04, 21.83it/s]
Loading 0:  75%|███████▌  | 273/363 [00:10<00:03, 28.31it/s]
Loading 0:  76%|███████▋  | 277/363 [00:10<00:03, 27.27it/s]
Loading 0:  78%|███████▊  | 282/363 [00:10<00:02, 29.79it/s]
Loading 0:  79%|███████▉  | 286/363 [00:10<00:02, 28.95it/s]
Loading 0:  80%|████████  | 291/363 [00:10<00:02, 31.80it/s]
Loading 0:  81%|████████▏ | 295/363 [00:10<00:02, 30.31it/s]
Loading 0:  82%|████████▏ | 299/363 [00:10<00:02, 29.48it/s]
Loading 0:  84%|████████▎ | 304/363 [00:11<00:02, 25.17it/s]
Loading 0:  85%|████████▍ | 307/363 [00:11<00:02, 23.62it/s]
Loading 0:  86%|████████▌ | 311/363 [00:11<00:02, 22.69it/s]
Loading 0:  88%|████████▊ | 318/363 [00:11<00:01, 29.26it/s]
Loading 0:  89%|████████▊ | 322/363 [00:11<00:01, 28.31it/s]
Loading 0:  90%|█████████ | 327/363 [00:11<00:01, 30.67it/s]
Loading 0:  91%|█████████ | 331/363 [00:12<00:01, 28.79it/s]
Loading 0:  93%|█████████▎| 336/363 [00:12<00:00, 30.71it/s]
Loading 0:  94%|█████████▎| 340/363 [00:12<00:00, 28.50it/s]
Loading 0:  95%|█████████▍| 344/363 [00:19<00:09,  1.98it/s]
Loading 0:  96%|█████████▌| 348/363 [00:19<00:05,  2.66it/s]
Loading 0:  97%|█████████▋| 353/363 [00:19<00:02,  3.86it/s]
Loading 0:  98%|█████████▊| 357/363 [00:19<00:01,  5.00it/s]
Loading 0: 100%|█████████▉| 362/363 [00:19<00:00,  7.10it/s]
                                                            
Job chaiml-albert-dpo-0912-v-2161-v1-mkmlizer completed after 168.19s with status: succeeded
Stopping job with name chaiml-albert-dpo-0912-v-2161-v1-mkmlizer
Pipeline stage MKMLizer completed in 169.24s
run pipeline stage %s
Running pipeline stage MKMLTemplater
Pipeline stage MKMLTemplater completed in 0.09s
run pipeline stage %s
Running pipeline stage MKMLDeployer
Creating inference service chaiml-albert-dpo-0912-v-2161-v1
Waiting for inference service chaiml-albert-dpo-0912-v-2161-v1 to be ready
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
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
Inference service chaiml-albert-dpo-0912-v-2161-v1 ready after 190.9113748073578s
Pipeline stage MKMLDeployer completed in 191.26s
run pipeline stage %s
Running pipeline stage StressChecker
Received healthy response to inference request in 1.7531869411468506s
Received healthy response to inference request in 1.641932487487793s
Received healthy response to inference request in 1.4160542488098145s
Received healthy response to inference request in 1.3234143257141113s
Received healthy response to inference request in 1.102813959121704s
5 requests
0 failed requests
5th percentile: 1.1469340324401855
10th percentile: 1.191054105758667
20th percentile: 1.2792942523956299
30th percentile: 1.341942310333252
40th percentile: 1.3789982795715332
50th percentile: 1.4160542488098145
60th percentile: 1.5064055442810058
70th percentile: 1.5967568397521972
80th percentile: 1.6641833782196045
90th percentile: 1.7086851596832275
95th percentile: 1.730936050415039
99th percentile: 1.7487367630004882
mean time: 1.4474803924560546
Pipeline stage StressChecker completed in 10.20s
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 6.54s
Shutdown handler de-registered
chaiml-albert-dpo-0912-v_2161_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 0.13s
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 chaiml-albert-dpo-0912-v-2161-v1-profiler
Waiting for inference service chaiml-albert-dpo-0912-v-2161-v1-profiler to be ready
Inference service chaiml-albert-dpo-0912-v-2161-v1-profiler ready after 180.4776635169983s
Pipeline stage MKMLProfilerDeployer completed in 180.89s
run pipeline stage %s
Running pipeline stage MKMLProfilerRunner
kubectl cp /code/guanaco/guanaco_inference_services/src/inference_scripts tenant-chaiml-guanaco/chaiml-albert-dpo-0920acbc437883e9ec91e4e5fd8f241172-deploklw59:/code/chaiverse_profiler_1726256047 --namespace tenant-chaiml-guanaco
kubectl exec -it chaiml-albert-dpo-0920acbc437883e9ec91e4e5fd8f241172-deploklw59 --namespace tenant-chaiml-guanaco -- sh -c 'cd /code/chaiverse_profiler_1726256047 && python profiles.py profile --best_of_n 8 --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 1024 --output_tokens 64 --summary /code/chaiverse_profiler_1726256047/summary.json'
kubectl exec -it chaiml-albert-dpo-0920acbc437883e9ec91e4e5fd8f241172-deploklw59 --namespace tenant-chaiml-guanaco -- bash -c 'cat /code/chaiverse_profiler_1726256047/summary.json'
Pipeline stage MKMLProfilerRunner completed in 1153.78s
run pipeline stage %s
Running pipeline stage MKMLProfilerDeleter
Checking if service chaiml-albert-dpo-0912-v-2161-v1-profiler is running
Tearing down inference service chaiml-albert-dpo-0912-v-2161-v1-profiler
Service chaiml-albert-dpo-0912-v-2161-v1-profiler has been torndown
Pipeline stage MKMLProfilerDeleter completed in 2.29s
Shutdown handler de-registered
chaiml-albert-dpo-0912-v_2161_v1 status is now inactive due to auto deactivation removed underperforming models