meta-llama-meta-llama-3-1-8b

submission_id: meta-llama-meta-llama-3-1-8b_v1

developer_uid: Meliodia

best_of: 8

celo_rating: 1213.09

display_name: meta318b

family_friendly_score: 0.0

formatter: {'memory_template': '', 'prompt_template': '', 'bot_template': '<|im_start|>assistant\n{bot_name}: {message}<|im_end|>\n', 'user_template': '<|im_start|>user\n{user_name}: {message}<|im_end|>\n', 'response_template': '<|im_start|>assistant\n{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': 1024, 'best_of': 8, 'max_output_tokens': 64}

gpu_counts: {'NVIDIA RTX A5000': 1}

is_internal_developer: True

language_model: meta-llama/Meta-Llama-3.1-8B

latencies: [{'batch_size': 1, 'throughput': 0.8682508939300785, 'latency_mean': 1.1516400897502899, 'latency_p50': 1.1514641046524048, 'latency_p90': 1.2664693355560304}, {'batch_size': 4, 'throughput': 1.8525274908415355, 'latency_mean': 2.1514269065856935, 'latency_p50': 2.1428385972976685, 'latency_p90': 2.3925614833831785}, {'batch_size': 5, 'throughput': 1.9983602484269634, 'latency_mean': 2.493928039073944, 'latency_p50': 2.5189926624298096, 'latency_p90': 2.80331654548645}, {'batch_size': 8, 'throughput': 2.1650233438697652, 'latency_mean': 3.668518785238266, 'latency_p50': 3.6676506996154785, 'latency_p90': 4.080697560310364}, {'batch_size': 10, 'throughput': 2.211447670826452, 'latency_mean': 4.484510614871978, 'latency_p50': 4.515714645385742, 'latency_p90': 5.009341192245484}, {'batch_size': 12, 'throughput': 2.200994369612197, 'latency_mean': 5.404622515439987, 'latency_p50': 5.444784045219421, 'latency_p90': 6.0696792840957645}, {'batch_size': 15, 'throughput': 2.1842714084751407, 'latency_mean': 6.786920663118362, 'latency_p50': 6.804333806037903, 'latency_p90': 7.6121653556823725}]

max_input_tokens: 1024

max_output_tokens: 64

model_architecture: LlamaForCausalLM

model_group: meta-llama/Meta-Llama-3.

model_name: meta318b

model_num_parameters: 8030261248.0

model_repo: meta-llama/Meta-Llama-3.1-8B

model_size: 8B

num_battles: 11132

num_wins: 4983

ranking_group: single

status: torndown

submission_type: basic

throughput_3p7s: 2.18

timestamp: 2024-09-19T22:31:24+00:00

us_pacific_date: 2024-09-19

win_ratio: 0.44762845849802374

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 meta-llama-meta-llama-3-1-8b-v1-mkmlizer
Waiting for job on meta-llama-meta-llama-3-1-8b-v1-mkmlizer to finish
Connection pool is full, discarding connection: %s. Connection pool size: %s
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ╔═════════════════════════════════════════════════════════════════════╗
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║     _____            __           __                                ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║    / _/ /_ ___    __/ /  ___ ___ / /                                ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║   / _/ / // / |/|/ / _ \/ -_) -_) /                                 ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║  /_//_/\_, /|__,__/_//_/\__/\__/_/                                  ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║       /___/                                                         ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║                                                                     ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║  Version: 0.10.1                                                    ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║  Copyright 2023 MK ONE TECHNOLOGIES Inc.                            ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║  https://mk1.ai                                                     ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║                                                                     ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║  The license key for the current software has been verified as      ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║  belonging to:                                                      ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║                                                                     ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║  Chai Research Corp.                                                ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║  Account ID: 7997a29f-0ceb-4cc7-9adf-840c57b4ae6f                   ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║  Expiration: 2024-10-15 23:59:59                                    ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ║                                                                     ║
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: ╚═════════════════════════════════════════════════════════════════════╝
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: Downloaded to shared memory in 38.225s
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: quantizing model to /dev/shm/model_cache, profile:s0, folder:/tmp/tmpoo9gauea, device:0
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: Saving flywheel model at /dev/shm/model_cache
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: quantized model in 25.368s
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: Processed model meta-llama/Meta-Llama-3.1-8B in 63.593s
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: creating bucket guanaco-mkml-models
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: Bucket 's3://guanaco-mkml-models/' created
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: uploading /dev/shm/model_cache to s3://guanaco-mkml-models/meta-llama-meta-llama-3-1-8b-v1
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: cp /dev/shm/model_cache/config.json s3://guanaco-mkml-models/meta-llama-meta-llama-3-1-8b-v1/config.json
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: cp /dev/shm/model_cache/special_tokens_map.json s3://guanaco-mkml-models/meta-llama-meta-llama-3-1-8b-v1/special_tokens_map.json
Failed to get response for submission chaiml-lexical-nemo-v4-1k1e5_v3: ('http://chaiml-llama-8b-pairwis-8189-v27-predictor.tenant-chaiml-guanaco.k2.chaiverse.com/v1/models/GPT-J-6B-lit-v2:predict', 'read tcp 127.0.0.1:50094->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
Connection pool is full, discarding connection: %s. Connection pool size: %s
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: cp /dev/shm/model_cache/tokenizer.json s3://guanaco-mkml-models/meta-llama-meta-llama-3-1-8b-v1/tokenizer.json
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: cp /dev/shm/model_cache/flywheel_model.0.safetensors s3://guanaco-mkml-models/meta-llama-meta-llama-3-1-8b-v1/flywheel_model.0.safetensors
meta-llama-meta-llama-3-1-8b-v1-mkmlizer: 
Loading 0:   0%|          | 0/291 [00:00<?, ?it/s]
Loading 0:   2%|▏         | 5/291 [00:00<00:07, 36.60it/s]
Loading 0:   5%|▍         | 14/291 [00:00<00:05, 51.26it/s]
Loading 0:   8%|▊         | 23/291 [00:00<00:04, 56.12it/s]
Loading 0:  11%|█         | 32/291 [00:00<00:04, 57.55it/s]
Loading 0:  14%|█▍        | 41/291 [00:00<00:04, 58.45it/s]
Loading 0:  17%|█▋        | 50/291 [00:00<00:04, 58.46it/s]
Loading 0:  20%|██        | 59/291 [00:01<00:03, 59.00it/s]
Loading 0:  23%|██▎       | 67/291 [00:01<00:03, 63.49it/s]
Loading 0:  25%|██▌       | 74/291 [00:01<00:03, 58.64it/s]
Loading 0:  27%|██▋       | 80/291 [00:01<00:03, 56.22it/s]
Loading 0:  30%|██▉       | 86/291 [00:01<00:05, 36.82it/s]
Loading 0:  33%|███▎      | 95/291 [00:01<00:04, 42.13it/s]
Loading 0:  36%|███▌      | 104/291 [00:02<00:04, 46.46it/s]
Loading 0:  39%|███▉      | 113/291 [00:02<00:03, 48.28it/s]
Loading 0:  42%|████▏     | 121/291 [00:02<00:03, 54.38it/s]
Loading 0:  44%|████▎     | 127/291 [00:02<00:03, 51.91it/s]
Loading 0:  46%|████▌     | 133/291 [00:02<00:03, 51.76it/s]
Loading 0:  48%|████▊     | 140/291 [00:02<00:03, 49.92it/s]
Loading 0:  51%|█████     | 149/291 [00:02<00:02, 52.24it/s]
Loading 0:  54%|█████▍    | 158/291 [00:03<00:02, 53.55it/s]
Loading 0:  57%|█████▋    | 167/291 [00:03<00:02, 54.04it/s]
Loading 0:  61%|██████    | 177/291 [00:03<00:01, 58.52it/s]
Loading 0:  63%|██████▎   | 183/291 [00:03<00:01, 57.53it/s]
Loading 0:  65%|██████▍   | 189/291 [00:03<00:02, 38.05it/s]
Loading 0:  67%|██████▋   | 194/291 [00:03<00:02, 39.62it/s]
Loading 0:  70%|██████▉   | 203/291 [00:04<00:01, 44.05it/s]
Loading 0:  73%|███████▎  | 212/291 [00:04<00:01, 47.81it/s]
Loading 0:  76%|███████▌  | 221/291 [00:04<00:01, 49.27it/s]
Loading 0:  79%|███████▊  | 229/291 [00:04<00:01, 55.16it/s]
Loading 0:  81%|████████  | 235/291 [00:04<00:01, 53.68it/s]
Loading 0:  83%|████████▎ | 241/291 [00:04<00:00, 52.55it/s]
Loading 0:  85%|████████▍ | 247/291 [00:04<00:00, 53.28it/s]
Loading 0:  87%|████████▋ | 253/291 [00:04<00:00, 49.10it/s]
Loading 0:  89%|████████▉ | 259/291 [00:05<00:00, 49.78it/s]
Loading 0:  91%|█████████ | 265/291 [00:05<00:00, 52.00it/s]
Loading 0:  93%|█████████▎| 271/291 [00:05<00:00, 50.15it/s]
Loading 0:  95%|█████████▌| 277/291 [00:05<00:00, 50.94it/s]
Loading 0:  97%|█████████▋| 283/291 [00:05<00:00, 47.61it/s]
Loading 0:  99%|█████████▉| 288/291 [00:10<00:00,  3.39it/s]
                                                            
Job meta-llama-meta-llama-3-1-8b-v1-mkmlizer completed after 83.62s with status: succeeded
Stopping job with name meta-llama-meta-llama-3-1-8b-v1-mkmlizer
Pipeline stage MKMLizer completed in 84.51s
run pipeline stage %s
Running pipeline stage MKMLTemplater
Pipeline stage MKMLTemplater completed in 0.11s
run pipeline stage %s
Running pipeline stage MKMLDeployer
Creating inference service meta-llama-meta-llama-3-1-8b-v1
Waiting for inference service meta-llama-meta-llama-3-1-8b-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
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
Failed to get response for submission chaiml-elo-alignment-run-3_v36: ('http://chaiml-elo-alignment-run-3-v36-predictor.tenant-chaiml-guanaco.k.chaiverse.com/v1/models/GPT-J-6B-lit-v2:predict', 'read tcp 127.0.0.1:42652->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
Connection pool is full, discarding connection: %s. Connection pool size: %s
Inference service meta-llama-meta-llama-3-1-8b-v1 ready after 201.7664930820465s
Pipeline stage MKMLDeployer completed in 202.15s
run pipeline stage %s
Running pipeline stage StressChecker
Received healthy response to inference request in 2.0245561599731445s
Received healthy response to inference request in 2.4245052337646484s
Received healthy response to inference request in 1.5720362663269043s
Received healthy response to inference request in 2.224947690963745s
Received healthy response to inference request in 1.386486530303955s
5 requests
0 failed requests
5th percentile: 1.423596477508545
10th percentile: 1.4607064247131347
20th percentile: 1.5349263191223144
30th percentile: 1.6625402450561524
40th percentile: 1.8435482025146486
50th percentile: 2.0245561599731445
60th percentile: 2.1047127723693846
70th percentile: 2.184869384765625
80th percentile: 2.264859199523926
90th percentile: 2.344682216644287
95th percentile: 2.3845937252044678
99th percentile: 2.4165229320526125
mean time: 1.9265063762664796
Pipeline stage StressChecker completed in 10.97s
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 6.03s
Shutdown handler de-registered
meta-llama-meta-llama-3-1-8b_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 0.12s
run pipeline stage %s
Running pipeline stage MKMLProfilerTemplater
Pipeline stage MKMLProfilerTemplater completed in 0.12s
run pipeline stage %s
Running pipeline stage MKMLProfilerDeployer
Creating inference service meta-llama-meta-llama-3-1-8b-v1-profiler
Waiting for inference service meta-llama-meta-llama-3-1-8b-v1-profiler to be ready
Inference service meta-llama-meta-llama-3-1-8b-v1-profiler ready after 200.44088125228882s
Pipeline stage MKMLProfilerDeployer completed in 200.82s
run pipeline stage %s
Running pipeline stage MKMLProfilerRunner
kubectl cp /code/guanaco/guanaco_inference_services/src/inference_scripts tenant-chaiml-guanaco/meta-llama-meta-llamd34f95e5f93999dcf75b70de388bf556-deplow6h4s:/code/chaiverse_profiler_1726785636 --namespace tenant-chaiml-guanaco
kubectl exec -it meta-llama-meta-llamd34f95e5f93999dcf75b70de388bf556-deplow6h4s --namespace tenant-chaiml-guanaco -- sh -c 'cd /code/chaiverse_profiler_1726785636 && 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_1726785636/summary.json'
kubectl exec -it meta-llama-meta-llamd34f95e5f93999dcf75b70de388bf556-deplow6h4s --namespace tenant-chaiml-guanaco -- bash -c 'cat /code/chaiverse_profiler_1726785636/summary.json'
Pipeline stage MKMLProfilerRunner completed in 808.92s
run pipeline stage %s
Running pipeline stage MKMLProfilerDeleter
Checking if service meta-llama-meta-llama-3-1-8b-v1-profiler is running
Tearing down inference service meta-llama-meta-llama-3-1-8b-v1-profiler
Service meta-llama-meta-llama-3-1-8b-v1-profiler has been torndown
Pipeline stage MKMLProfilerDeleter completed in 2.21s
Shutdown handler de-registered
meta-llama-meta-llama-3-1-8b_v1 status is now inactive due to auto deactivation removed underperforming models
meta-llama-meta-llama-3-1-8b_v1 status is now torndown due to DeploymentManager action