submission_id: nousresearch-meta-llama-3-8b_v23
developer_uid: chai_backend_admin
alignment_samples: 11479
alignment_score: -0.8589611619422162
best_of: 2
celo_rating: 1154.09
display_name: nousresearch-meta-llama-3-8b_v23
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': 0.99, 'min_p': 0.1, 'top_k': 40, 'presence_penalty': 0.0, 'frequency_penalty': 0.0, 'stopping_words': ['\n'], 'max_input_tokens': 512, 'best_of': 2, 'max_output_tokens': 64}
gpu_counts: {'NVIDIA RTX A5000': 1}
is_internal_developer: True
language_model: NousResearch/Meta-Llama-3-8B
latencies: [{'batch_size': 1, 'throughput': 1.0369168153023727, 'latency_mean': 0.9643385553359985, 'latency_p50': 0.9611647129058838, 'latency_p90': 1.0779971361160279}, {'batch_size': 5, 'throughput': 3.337804452211329, 'latency_mean': 1.482992125749588, 'latency_p50': 1.4641920328140259, 'latency_p90': 1.686904287338257}, {'batch_size': 10, 'throughput': 4.774166234060552, 'latency_mean': 2.076084454059601, 'latency_p50': 2.0917528867721558, 'latency_p90': 2.34508376121521}, {'batch_size': 15, 'throughput': 5.472508295031052, 'latency_mean': 2.707471194267273, 'latency_p50': 2.7371010780334473, 'latency_p90': 3.0606255292892457}, {'batch_size': 20, 'throughput': 5.984771117110906, 'latency_mean': 3.2963728034496307, 'latency_p50': 3.2832378149032593, 'latency_p90': 3.7705803632736203}, {'batch_size': 25, 'throughput': 6.141391842645459, 'latency_mean': 3.9879662477970124, 'latency_p50': 4.011241555213928, 'latency_p90': 4.585857462882996}, {'batch_size': 30, 'throughput': 6.258047709559781, 'latency_mean': 4.67686963558197, 'latency_p50': 4.68639862537384, 'latency_p90': 5.378696799278259}, {'batch_size': 35, 'throughput': 6.2673470045683395, 'latency_mean': 5.419337519407272, 'latency_p50': 5.4146119356155396, 'latency_p90': 6.169280791282653}]
max_input_tokens: 512
max_output_tokens: 64
model_architecture: LlamaForCausalLM
model_group: NousResearch/Meta-Llama-
model_name: nousresearch-meta-llama-3-8b_v23
model_num_parameters: 8030261248.0
model_repo: NousResearch/Meta-Llama-3-8B
model_size: 8B
num_battles: 11479
num_wins: 4406
propriety_score: 0.7215686274509804
propriety_total_count: 1020.0
ranking_group: single
status: inactive
submission_type: basic
throughput_3p7s: 6.19
timestamp: 2024-09-07T03:05:34+00:00
us_pacific_date: 2024-09-06
win_ratio: 0.3838313441937451
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 nousresearch-meta-llama-3-8b-v23-mkmlizer
Waiting for job on nousresearch-meta-llama-3-8b-v23-mkmlizer to finish
nousresearch-meta-llama-3-8b-v23-mkmlizer: ╔═════════════════════════════════════════════════════════════════════╗
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║     _____            __           __                                ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║    / _/ /_ ___    __/ /  ___ ___ / /                                ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║   / _/ / // / |/|/ / _ \/ -_) -_) /                                 ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║  /_//_/\_, /|__,__/_//_/\__/\__/_/                                  ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║       /___/                                                         ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║                                                                     ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║  Version: 0.10.1                                                    ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║  Copyright 2023 MK ONE TECHNOLOGIES Inc.                            ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║  https://mk1.ai                                                     ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║                                                                     ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║  The license key for the current software has been verified as      ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║  belonging to:                                                      ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║                                                                     ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║  Chai Research Corp.                                                ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║  Account ID: 7997a29f-0ceb-4cc7-9adf-840c57b4ae6f                   ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║  Expiration: 2024-10-15 23:59:59                                    ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ║                                                                     ║
nousresearch-meta-llama-3-8b-v23-mkmlizer: ╚═════════════════════════════════════════════════════════════════════╝
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
nousresearch-meta-llama-3-8b-v23-mkmlizer: Downloaded to shared memory in 25.464s
nousresearch-meta-llama-3-8b-v23-mkmlizer: quantizing model to /dev/shm/model_cache, profile:s0, folder:/tmp/tmp55e3i9x6, device:0
nousresearch-meta-llama-3-8b-v23-mkmlizer: Saving flywheel model at /dev/shm/model_cache
nousresearch-meta-llama-3-8b-v23-mkmlizer: quantized model in 26.592s
nousresearch-meta-llama-3-8b-v23-mkmlizer: Processed model NousResearch/Meta-Llama-3-8B in 52.057s
nousresearch-meta-llama-3-8b-v23-mkmlizer: creating bucket guanaco-mkml-models
nousresearch-meta-llama-3-8b-v23-mkmlizer: Bucket 's3://guanaco-mkml-models/' created
nousresearch-meta-llama-3-8b-v23-mkmlizer: uploading /dev/shm/model_cache to s3://guanaco-mkml-models/nousresearch-meta-llama-3-8b-v23
nousresearch-meta-llama-3-8b-v23-mkmlizer: cp /dev/shm/model_cache/config.json s3://guanaco-mkml-models/nousresearch-meta-llama-3-8b-v23/config.json
nousresearch-meta-llama-3-8b-v23-mkmlizer: cp /dev/shm/model_cache/special_tokens_map.json s3://guanaco-mkml-models/nousresearch-meta-llama-3-8b-v23/special_tokens_map.json
nousresearch-meta-llama-3-8b-v23-mkmlizer: cp /dev/shm/model_cache/tokenizer_config.json s3://guanaco-mkml-models/nousresearch-meta-llama-3-8b-v23/tokenizer_config.json
nousresearch-meta-llama-3-8b-v23-mkmlizer: cp /dev/shm/model_cache/tokenizer.json s3://guanaco-mkml-models/nousresearch-meta-llama-3-8b-v23/tokenizer.json
nousresearch-meta-llama-3-8b-v23-mkmlizer: cp /dev/shm/model_cache/flywheel_model.0.safetensors s3://guanaco-mkml-models/nousresearch-meta-llama-3-8b-v23/flywheel_model.0.safetensors
nousresearch-meta-llama-3-8b-v23-mkmlizer: 
Loading 0:   0%|          | 0/291 [00:00<?, ?it/s]
Loading 0:   2%|▏         | 5/291 [00:00<00:08, 33.22it/s]
Loading 0:   4%|▍         | 13/291 [00:00<00:05, 55.05it/s]
Loading 0:   7%|▋         | 19/291 [00:00<00:05, 51.36it/s]
Loading 0:   9%|▊         | 25/291 [00:00<00:05, 51.50it/s]
Loading 0:  11%|█         | 31/291 [00:00<00:04, 52.15it/s]
Loading 0:  13%|█▎        | 37/291 [00:00<00:05, 47.96it/s]
Loading 0:  14%|█▍        | 42/291 [00:00<00:05, 46.89it/s]
Loading 0:  17%|█▋        | 49/291 [00:00<00:04, 53.09it/s]
Loading 0:  19%|█▉        | 55/291 [00:01<00:05, 46.47it/s]
Loading 0:  21%|██        | 60/291 [00:01<00:05, 44.95it/s]
Loading 0:  23%|██▎       | 68/291 [00:01<00:04, 46.20it/s]
Loading 0:  26%|██▌       | 76/291 [00:01<00:03, 54.15it/s]
Loading 0:  28%|██▊       | 82/291 [00:01<00:04, 50.10it/s]
Loading 0:  30%|███       | 88/291 [00:01<00:05, 35.21it/s]
Loading 0:  32%|███▏      | 93/291 [00:02<00:05, 37.49it/s]
Loading 0:  34%|███▎      | 98/291 [00:02<00:04, 39.68it/s]
Loading 0:  36%|███▌      | 104/291 [00:02<00:05, 37.08it/s]
Loading 0:  38%|███▊      | 111/291 [00:02<00:04, 43.25it/s]
Loading 0:  40%|███▉      | 116/291 [00:02<00:03, 44.07it/s]
Loading 0:  42%|████▏     | 121/291 [00:02<00:03, 45.20it/s]
Loading 0:  44%|████▎     | 127/291 [00:02<00:03, 41.82it/s]
Loading 0:  45%|████▌     | 132/291 [00:02<00:03, 40.66it/s]
Loading 0:  48%|████▊     | 139/291 [00:03<00:03, 47.32it/s]
Loading 0:  50%|████▉     | 145/291 [00:03<00:03, 45.34it/s]
Loading 0:  52%|█████▏    | 150/291 [00:03<00:03, 45.02it/s]
Loading 0:  54%|█████▍    | 157/291 [00:03<00:02, 49.92it/s]
Loading 0:  56%|█████▌    | 163/291 [00:03<00:02, 43.98it/s]
Loading 0:  58%|█████▊    | 168/291 [00:03<00:02, 43.60it/s]
Loading 0:  59%|█████▉    | 173/291 [00:03<00:02, 44.03it/s]
Loading 0:  62%|██████▏   | 180/291 [00:03<00:02, 48.57it/s]
Loading 0:  64%|██████▍   | 186/291 [00:04<00:02, 45.17it/s]
Loading 0:  66%|██████▌   | 191/291 [00:04<00:03, 32.15it/s]
Loading 0:  67%|██████▋   | 195/291 [00:04<00:02, 32.36it/s]
Loading 0:  69%|██████▉   | 202/291 [00:04<00:02, 39.66it/s]
Loading 0:  71%|███████▏  | 208/291 [00:04<00:02, 38.11it/s]
Loading 0:  73%|███████▎  | 213/291 [00:04<00:01, 39.18it/s]
Loading 0:  75%|███████▌  | 219/291 [00:05<00:01, 43.32it/s]
Loading 0:  77%|███████▋  | 224/291 [00:05<00:01, 43.71it/s]
Loading 0:  79%|███████▊  | 229/291 [00:05<00:01, 43.52it/s]
Loading 0:  80%|████████  | 234/291 [00:05<00:01, 44.85it/s]
Loading 0:  82%|████████▏ | 239/291 [00:05<00:01, 38.43it/s]
Loading 0:  85%|████████▍ | 246/291 [00:05<00:00, 45.59it/s]
Loading 0:  86%|████████▋ | 251/291 [00:05<00:00, 44.80it/s]
Loading 0:  88%|████████▊ | 256/291 [00:05<00:00, 45.57it/s]
Loading 0:  90%|████████▉ | 261/291 [00:05<00:00, 46.00it/s]
Loading 0:  91%|█████████▏| 266/291 [00:06<00:00, 37.16it/s]
Loading 0:  94%|█████████▍| 273/291 [00:06<00:00, 43.98it/s]
Loading 0:  96%|█████████▌| 278/291 [00:06<00:00, 44.39it/s]
Loading 0:  97%|█████████▋| 283/291 [00:06<00:00, 38.74it/s]
Loading 0:  99%|█████████▉| 288/291 [00:11<00:00,  3.07it/s]
                                                            

Job nousresearch-meta-llama-3-8b-v23-mkmlizer completed after 125.78s with status: succeeded
Stopping job with name nousresearch-meta-llama-3-8b-v23-mkmlizer
Pipeline stage MKMLizer completed in 126.74s
run pipeline stage %s
Running pipeline stage MKMLTemplater
Pipeline stage MKMLTemplater completed in 0.08s
run pipeline stage %s
Running pipeline stage MKMLDeployer
Creating inference service nousresearch-meta-llama-3-8b-v23
Waiting for inference service nousresearch-meta-llama-3-8b-v23 to be ready
Retrying (%r) after connection broken by '%r': %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 nousresearch-meta-llama-3-8b-v23 ready after 151.20497679710388s
Pipeline stage MKMLDeployer completed in 152.05s
run pipeline stage %s
Running pipeline stage StressChecker
Received healthy response to inference request in 2.0089285373687744s
Received healthy response to inference request in 1.7835235595703125s
Received healthy response to inference request in 2.308650016784668s
Received healthy response to inference request in 1.3797686100006104s
Received healthy response to inference request in 1.0122950077056885s
5 requests
0 failed requests
5th percentile: 1.085789728164673
10th percentile: 1.1592844486236573
20th percentile: 1.306273889541626
30th percentile: 1.4605195999145508
40th percentile: 1.6220215797424316
50th percentile: 1.7835235595703125
60th percentile: 1.8736855506896972
70th percentile: 1.963847541809082
80th percentile: 2.068872833251953
90th percentile: 2.1887614250183107
95th percentile: 2.2487057209014893
99th percentile: 2.2966611576080322
mean time: 1.6986331462860107
Pipeline stage StressChecker completed in 10.33s
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 4.69s
Shutdown handler de-registered
nousresearch-meta-llama-3-8b_v23 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 0.11s
run pipeline stage %s
Running pipeline stage MKMLProfilerDeployer
Creating inference service nousresearch-meta-llama-3-8b-v23-profiler
Waiting for inference service nousresearch-meta-llama-3-8b-v23-profiler to be ready
Inference service nousresearch-meta-llama-3-8b-v23-profiler ready after 150.3668999671936s
Pipeline stage MKMLProfilerDeployer completed in 150.72s
run pipeline stage %s
Running pipeline stage MKMLProfilerRunner
kubectl cp /code/guanaco/guanaco_inference_services/src/inference_scripts tenant-chaiml-guanaco/nousresearch-meta-ll4a52be0e13abb12c01ba91cf625d56a6-deploswprv:/code/chaiverse_profiler_1725678817 --namespace tenant-chaiml-guanaco
kubectl exec -it nousresearch-meta-ll4a52be0e13abb12c01ba91cf625d56a6-deploswprv --namespace tenant-chaiml-guanaco -- sh -c 'cd /code/chaiverse_profiler_1725678817 && python profiles.py profile --best_of_n 2 --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_1725678817/summary.json'
kubectl exec -it nousresearch-meta-ll4a52be0e13abb12c01ba91cf625d56a6-deploswprv --namespace tenant-chaiml-guanaco -- bash -c 'cat /code/chaiverse_profiler_1725678817/summary.json'
Pipeline stage MKMLProfilerRunner completed in 466.77s
run pipeline stage %s
Running pipeline stage MKMLProfilerDeleter
Checking if service nousresearch-meta-llama-3-8b-v23-profiler is running
Tearing down inference service nousresearch-meta-llama-3-8b-v23-profiler
Service nousresearch-meta-llama-3-8b-v23-profiler has been torndown
Pipeline stage MKMLProfilerDeleter completed in 1.74s
Shutdown handler de-registered
nousresearch-meta-llama-3-8b_v23 status is now inactive due to auto deactivation removed underperforming models

Usage Metrics

Latency Metrics