chaiml-gemma3-27b-sft-exp1

developer_uid: NischayDnk

submission_id: chaiml-gemma3-27b-sft-exp1_v1

model_name: chaiml-gemma3-27b-sft-exp1_v1

model_group: ChaiML/gemma3-27b-sft-ex

status: torndown

timestamp: 2025-04-24T03:45:34+00:00

num_battles: 7606

num_wins: 3519

celo_rating: 1252.97

family_friendly_score: 0.5429999999999999

family_friendly_standard_error: 0.007044870474323854

submission_type: basic

model_repo: ChaiML/gemma3-27b-sft-exp1

model_architecture: Gemma3ForCausalLM

model_num_parameters: 29698212096.0

best_of: 8

max_input_tokens: 512

max_output_tokens: 64

reward_model: default

latencies: [{'batch_size': 1, 'throughput': 0.2957306083389341, 'latency_mean': 3.3813996076583863, 'latency_p50': 3.319413900375366, 'latency_p90': 3.8240628957748415}, {'batch_size': 2, 'throughput': 0.4977815617933292, 'latency_mean': 4.014540853500367, 'latency_p50': 4.016810655593872, 'latency_p90': 4.437553739547729}, {'batch_size': 3, 'throughput': 0.696420158310648, 'latency_mean': 4.294725053310394, 'latency_p50': 4.245514392852783, 'latency_p90': 4.802515721321106}, {'batch_size': 4, 'throughput': 0.8497203690697331, 'latency_mean': 4.689823083877563, 'latency_p50': 4.705562710762024, 'latency_p90': 5.27860107421875}, {'batch_size': 5, 'throughput': 0.9674454981080869, 'latency_mean': 5.137126441001892, 'latency_p50': 5.160694122314453, 'latency_p90': 5.6846835851669315}]

gpu_counts: {'NVIDIA A100-SXM4-80GB': 1}

display_name: chaiml-gemma3-27b-sft-exp1_v1

is_internal_developer: False

language_model: ChaiML/gemma3-27b-sft-exp1

model_size: 30B

ranking_group: single

throughput_3p7s: 0.41

us_pacific_date: 2025-04-23

win_ratio: 0.4626610570602156

generation_params: {'temperature': 0.45, 'top_p': 0.95, 'min_p': 0.025, 'top_k': 40, 'presence_penalty': 0.4, 'frequency_penalty': 0.4, 'stopping_words': ['<|im_end|>', '<|im_start|>', '\n'], 'max_input_tokens': 512, 'best_of': 8, 'max_output_tokens': 64}

formatter: {'memory_template': '<|system|>Family Friendly{memory}\n', 'prompt_template': '', 'bot_template': '<|im_start|>assistant\n{message}<|im_end|>\n', 'user_template': '<|im_start|>user\nYou:{message}<|im_end|>\n', 'response_template': '<|im_start|>assistant\n', 'truncate_by_message': False}

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-gemma3-27b-sft-exp1-v1-mkmlizer
Waiting for job on chaiml-gemma3-27b-sft-exp1-v1-mkmlizer to finish
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ╔═════════════════════════════════════════════════════════════════════╗
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║                                                                     ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║   ██████   ██████  █████   ████  ████                               ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║  ░░██████ ██████  ░░███   ███░  ░░███                               ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║   ░███░█████░███   ░███  ███     ░███                               ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║   ░███░░███ ░███   ░███████      ░███                               ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║   ░███ ░░░  ░███   ░███░░███     ░███                               ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║   ░███      ░███   ░███ ░░███    ░███                               ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║   █████     █████  █████ ░░████  █████                              ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║  ░░░░░     ░░░░░  ░░░░░   ░░░░  ░░░░░                               ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║                                                                     ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║  Version: 0.28.3                                                    ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║  Features: FLYWHEEL, CUDA                                           ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║  Copyright 2023-2025 MK ONE TECHNOLOGIES Inc.                       ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║  https://mk1.ai                                                     ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║                                                                     ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║  The license key for the current software has been verified as      ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║  belonging to:                                                      ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║                                                                     ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║  Chai Research Corp.                                                ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║  Account ID: 7997a29f-0ceb-4cc7-9adf-840c57b4ae6f                   ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║  Expiration: 2028-03-31 23:59:59                                    ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ║                                                                     ║
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: ╚═════════════════════════════════════════════════════════════════════╝
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: Downloaded to shared memory in 92.457s
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: quantizing model to /dev/shm/model_cache, profile:s0, folder:/tmp/tmp_3ya1pgo, device:0
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: Saving flywheel model at /dev/shm/model_cache
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: quantized model in 62.163s
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: Processed model ChaiML/gemma3-27b-sft-exp1 in 154.620s
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: creating bucket guanaco-mkml-models
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: Bucket 's3://guanaco-mkml-models/' created
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: uploading /dev/shm/model_cache to s3://guanaco-mkml-models/chaiml-gemma3-27b-sft-exp1-v1
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: cp /dev/shm/model_cache/config.json s3://guanaco-mkml-models/chaiml-gemma3-27b-sft-exp1-v1/config.json
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: cp /dev/shm/model_cache/added_tokens.json s3://guanaco-mkml-models/chaiml-gemma3-27b-sft-exp1-v1/added_tokens.json
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: cp /dev/shm/model_cache/special_tokens_map.json s3://guanaco-mkml-models/chaiml-gemma3-27b-sft-exp1-v1/special_tokens_map.json
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: cp /dev/shm/model_cache/tokenizer_config.json s3://guanaco-mkml-models/chaiml-gemma3-27b-sft-exp1-v1/tokenizer_config.json
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: cp /dev/shm/model_cache/tokenizer.model s3://guanaco-mkml-models/chaiml-gemma3-27b-sft-exp1-v1/tokenizer.model
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: cp /dev/shm/model_cache/tokenizer.json s3://guanaco-mkml-models/chaiml-gemma3-27b-sft-exp1-v1/tokenizer.json
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: cp /dev/shm/model_cache/flywheel_model.2.safetensors s3://guanaco-mkml-models/chaiml-gemma3-27b-sft-exp1-v1/flywheel_model.2.safetensors
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: cp /dev/shm/model_cache/flywheel_model.1.safetensors s3://guanaco-mkml-models/chaiml-gemma3-27b-sft-exp1-v1/flywheel_model.1.safetensors
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: cp /dev/shm/model_cache/flywheel_model.0.safetensors s3://guanaco-mkml-models/chaiml-gemma3-27b-sft-exp1-v1/flywheel_model.0.safetensors
chaiml-gemma3-27b-sft-exp1-v1-mkmlizer: 
Loading 0:   0%|          | 0/808 [00:00<?, ?it/s]
Loading 0:   1%|          | 5/808 [00:00<00:16, 49.44it/s]
Loading 0:   2%|▏         | 14/808 [00:00<00:11, 66.91it/s]
Loading 0:   3%|▎         | 21/808 [00:00<00:17, 45.15it/s]
Loading 0:   3%|▎         | 28/808 [00:00<00:16, 46.72it/s]
Loading 0:   4%|▍         | 35/808 [00:00<00:19, 38.90it/s]
Loading 0:   5%|▌         | 42/808 [00:00<00:16, 45.65it/s]
Loading 0:   7%|▋         | 53/808 [00:01<00:14, 53.27it/s]
Loading 0:   7%|▋         | 59/808 [00:01<00:14, 52.77it/s]
Loading 0:   8%|▊         | 68/808 [00:01<00:12, 57.16it/s]
Loading 0:  10%|▉         | 79/808 [00:01<00:12, 60.36it/s]
Loading 0:  11%|█         | 86/808 [00:01<00:12, 59.46it/s]
Loading 0:  12%|█▏        | 94/808 [00:01<00:12, 59.32it/s]
Loading 0:  13%|█▎        | 105/808 [00:01<00:11, 63.17it/s]
Loading 0:  14%|█▍        | 112/808 [00:02<00:11, 60.57it/s]
Loading 0:  15%|█▍        | 119/808 [00:02<00:15, 44.84it/s]
Loading 0:  16%|█▌        | 127/808 [00:02<00:14, 48.35it/s]
Loading 0:  17%|█▋        | 138/808 [00:02<00:12, 55.40it/s]
Loading 0:  18%|█▊        | 144/808 [00:02<00:12, 54.32it/s]
Loading 0:  19%|█▉        | 153/808 [00:02<00:11, 57.64it/s]
Loading 0:  20%|██        | 164/808 [00:02<00:10, 62.07it/s]
Loading 0:  21%|██        | 171/808 [00:03<00:10, 61.26it/s]
Loading 0:  22%|██▏       | 179/808 [00:03<00:10, 62.57it/s]
Loading 0:  24%|██▎       | 190/808 [00:03<00:09, 65.34it/s]
Loading 0:  24%|██▍       | 197/808 [00:03<00:12, 50.81it/s]
Loading 0:  25%|██▌       | 203/808 [00:03<00:11, 52.29it/s]
Loading 0:  26%|██▌       | 209/808 [00:03<00:11, 52.95it/s]
Loading 0:  27%|██▋       | 215/808 [00:03<00:11, 50.53it/s]
Loading 0:  28%|██▊       | 224/808 [00:04<00:10, 55.84it/s]
Loading 0:  29%|██▉       | 235/808 [00:04<00:09, 62.52it/s]
Loading 0:  30%|██▉       | 242/808 [00:04<00:09, 61.80it/s]
Loading 0:  31%|███       | 250/808 [00:04<00:09, 61.94it/s]
Loading 0:  32%|███▏      | 261/808 [00:04<00:08, 63.42it/s]
Loading 0:  33%|███▎      | 268/808 [00:04<00:09, 57.90it/s]
Loading 0:  34%|███▍      | 274/808 [00:05<00:12, 43.34it/s]
Loading 0:  35%|███▍      | 279/808 [00:05<00:11, 44.26it/s]
Loading 0:  36%|███▌      | 289/808 [00:05<00:10, 50.43it/s]
Loading 0:  37%|███▋      | 300/808 [00:05<00:09, 54.79it/s]
Loading 0:  37%|███▋      | 301/808 [00:20<00:09, 54.79it/s]
Loading 0:  37%|███▋      | 302/808 [00:20<05:41,  1.48it/s]
Loading 0:  39%|███▊      | 313/808 [00:20<03:19,  2.48it/s]
Loading 0:  39%|███▉      | 319/808 [00:20<02:35,  3.14it/s]
Loading 0:  41%|████      | 328/808 [00:20<01:43,  4.65it/s]
Loading 0:  42%|████▏     | 339/808 [00:21<01:05,  7.14it/s]
Loading 0:  43%|████▎     | 346/808 [00:21<00:50,  9.13it/s]
Loading 0:  44%|████▎     | 352/808 [00:21<00:42, 10.79it/s]
Loading 0:  44%|████▍     | 357/808 [00:21<00:34, 13.01it/s]
Loading 0:  45%|████▌     | 367/808 [00:21<00:23, 18.99it/s]
Loading 0:  47%|████▋     | 378/808 [00:21<00:16, 26.07it/s]
Loading 0:  48%|████▊     | 384/808 [00:21<00:14, 29.40it/s]
Loading 0:  49%|████▊     | 393/808 [00:22<00:11, 36.20it/s]
Loading 0:  50%|█████     | 404/808 [00:22<00:09, 44.66it/s]
Loading 0:  51%|█████     | 411/808 [00:22<00:08, 47.43it/s]
Loading 0:  52%|█████▏    | 418/808 [00:22<00:07, 49.11it/s]
Loading 0:  53%|█████▎    | 425/808 [00:22<00:09, 41.22it/s]
Loading 0:  53%|█████▎    | 432/808 [00:22<00:08, 45.19it/s]
Loading 0:  55%|█████▍    | 443/808 [00:23<00:07, 51.63it/s]
Loading 0:  56%|█████▌    | 449/808 [00:23<00:07, 48.92it/s]
Loading 0:  57%|█████▋    | 458/808 [00:23<00:06, 54.38it/s]
Loading 0:  58%|█████▊    | 469/808 [00:23<00:05, 58.98it/s]
Loading 0:  59%|█████▉    | 476/808 [00:23<00:05, 58.77it/s]
Loading 0:  60%|█████▉    | 484/808 [00:23<00:05, 58.86it/s]
Loading 0:  61%|██████▏   | 495/808 [00:23<00:05, 61.46it/s]
Loading 0:  62%|██████▏   | 502/808 [00:23<00:05, 59.02it/s]
Loading 0:  63%|██████▎   | 508/808 [00:24<00:06, 44.65it/s]
Loading 0:  63%|██████▎   | 513/808 [00:24<00:06, 45.13it/s]
Loading 0:  65%|██████▍   | 523/808 [00:24<00:05, 52.09it/s]
Loading 0:  66%|██████▌   | 534/808 [00:24<00:04, 57.12it/s]
Loading 0:  67%|██████▋   | 540/808 [00:24<00:04, 54.99it/s]
Loading 0:  68%|██████▊   | 549/808 [00:24<00:04, 57.07it/s]
Loading 0:  69%|██████▉   | 560/808 [00:25<00:04, 58.81it/s]
Loading 0:  70%|███████   | 566/808 [00:25<00:04, 57.08it/s]
Loading 0:  71%|███████   | 574/808 [00:25<00:04, 57.29it/s]
Loading 0:  72%|███████▏  | 581/808 [00:25<00:05, 44.71it/s]
Loading 0:  73%|███████▎  | 588/808 [00:25<00:04, 49.12it/s]
Loading 0:  74%|███████▍  | 599/808 [00:25<00:03, 54.95it/s]
Loading 0:  75%|███████▍  | 605/808 [00:25<00:03, 52.29it/s]
Loading 0:  76%|███████▌  | 614/808 [00:26<00:03, 55.54it/s]
Loading 0:  77%|███████▋  | 625/808 [00:26<00:03, 59.35it/s]
Loading 0:  78%|███████▊  | 632/808 [00:26<00:02, 59.38it/s]
Loading 0:  79%|███████▉  | 640/808 [00:26<00:02, 59.23it/s]
Loading 0:  81%|████████  | 651/808 [00:26<00:02, 61.98it/s]
Loading 0:  81%|████████▏ | 658/808 [00:26<00:02, 58.38it/s]
Loading 0:  82%|████████▏ | 664/808 [00:27<00:03, 44.41it/s]
Loading 0:  83%|████████▎ | 669/808 [00:27<00:03, 45.08it/s]
Loading 0:  84%|████████▍ | 679/808 [00:27<00:02, 52.10it/s]
Loading 0:  85%|████████▌ | 690/808 [00:27<00:02, 55.79it/s]
Loading 0:  86%|████████▌ | 696/808 [00:27<00:02, 54.36it/s]
Loading 0:  87%|████████▋ | 705/808 [00:27<00:01, 58.16it/s]
Loading 0:  89%|████████▊ | 716/808 [00:27<00:01, 62.97it/s]
Loading 0:  89%|████████▊ | 717/808 [00:42<00:01, 62.97it/s]
Loading 0:  89%|████████▉ | 718/808 [00:42<00:58,  1.54it/s]
Loading 0:  90%|█████████ | 729/808 [00:42<00:31,  2.53it/s]
Loading 0:  91%|█████████ | 735/808 [00:43<00:22,  3.24it/s]
Loading 0:  92%|█████████▏| 740/808 [00:43<00:17,  3.93it/s]
Loading 0:  92%|█████████▏| 746/808 [00:43<00:11,  5.29it/s]
Loading 0:  94%|█████████▎| 757/808 [00:43<00:05,  8.69it/s]
Loading 0:  95%|█████████▌| 768/808 [00:43<00:03, 12.88it/s]
Loading 0:  96%|█████████▌| 774/808 [00:43<00:02, 15.51it/s]
Loading 0:  97%|█████████▋| 783/808 [00:44<00:01, 20.53it/s]
Loading 0:  98%|█████████▊| 794/808 [00:44<00:00, 27.52it/s]
Loading 0:  99%|█████████▉| 801/808 [00:44<00:00, 31.79it/s]
                                                            
Job chaiml-gemma3-27b-sft-exp1-v1-mkmlizer completed after 238.53s with status: succeeded
Stopping job with name chaiml-gemma3-27b-sft-exp1-v1-mkmlizer
Pipeline stage MKMLizer completed in 239.01s
run pipeline stage %s
Running pipeline stage MKMLTemplater
Pipeline stage MKMLTemplater completed in 0.15s
run pipeline stage %s
Running pipeline stage MKMLDeployer
Creating inference service chaiml-gemma3-27b-sft-exp1-v1
Waiting for inference service chaiml-gemma3-27b-sft-exp1-v1 to be ready
Inference service chaiml-gemma3-27b-sft-exp1-v1 ready after 140.52102732658386s
Pipeline stage MKMLDeployer completed in 141.02s
run pipeline stage %s
Running pipeline stage StressChecker
Received healthy response to inference request in 4.373780012130737s
Received healthy response to inference request in 3.936511516571045s
Received healthy response to inference request in 3.715083122253418s
Received healthy response to inference request in 3.4511914253234863s
Received healthy response to inference request in 3.551368474960327s
5 requests
0 failed requests
5th percentile: 3.4712268352508544
10th percentile: 3.4912622451782225
20th percentile: 3.531333065032959
30th percentile: 3.5841114044189455
40th percentile: 3.6495972633361817
50th percentile: 3.715083122253418
60th percentile: 3.8036544799804686
70th percentile: 3.8922258377075196
80th percentile: 4.023965215682983
90th percentile: 4.19887261390686
95th percentile: 4.286326313018799
99th percentile: 4.3562892723083495
mean time: 3.805586910247803
%s, retrying in %s seconds...
Received healthy response to inference request in 3.476405382156372s
Received healthy response to inference request in 3.8486289978027344s
Received healthy response to inference request in 3.4904708862304688s
Received healthy response to inference request in 3.6712563037872314s
Received healthy response to inference request in 3.838578939437866s
5 requests
0 failed requests
5th percentile: 3.4792184829711914
10th percentile: 3.4820315837860107
20th percentile: 3.4876577854156494
30th percentile: 3.5266279697418215
40th percentile: 3.5989421367645265
50th percentile: 3.6712563037872314
60th percentile: 3.7381853580474855
70th percentile: 3.805114412307739
80th percentile: 3.8405889511108398
90th percentile: 3.8446089744567873
95th percentile: 3.846618986129761
99th percentile: 3.84822699546814
mean time: 3.6650681018829347
%s, retrying in %s seconds...
Received healthy response to inference request in 3.665419101715088s
Received healthy response to inference request in 3.6012279987335205s
Received healthy response to inference request in 3.679722785949707s
Received healthy response to inference request in 3.5607452392578125s
Received healthy response to inference request in 3.7242372035980225s
5 requests
0 failed requests
5th percentile: 3.568841791152954
10th percentile: 3.576938343048096
20th percentile: 3.593131446838379
30th percentile: 3.614066219329834
40th percentile: 3.6397426605224608
50th percentile: 3.665419101715088
60th percentile: 3.6711405754089355
70th percentile: 3.676862049102783
80th percentile: 3.68862566947937
90th percentile: 3.706431436538696
95th percentile: 3.7153343200683593
99th percentile: 3.72245662689209
mean time: 3.6462704658508303
Pipeline stage StressChecker completed in 59.73s
run pipeline stage %s
Running pipeline stage OfflineFamilyFriendlyTriggerPipeline
run_pipeline:run_in_cloud %s
starting trigger_guanaco_pipeline args=%s
triggered trigger_guanaco_pipeline args=%s
Pipeline stage OfflineFamilyFriendlyTriggerPipeline completed in 0.85s
run pipeline stage %s
Running pipeline stage TriggerMKMLProfilingPipeline
run_pipeline:run_in_cloud %s
starting trigger_guanaco_pipeline args=%s
triggered trigger_guanaco_pipeline args=%s
Pipeline stage TriggerMKMLProfilingPipeline completed in 0.69s
Shutdown handler de-registered
chaiml-gemma3-27b-sft-exp1_v1 status is now deployed due to DeploymentManager action
Shutdown handler registered
run pipeline %s
run pipeline stage %s
Running pipeline stage OfflineFamilyFriendlyScorer
Evaluating %s Family Friendly Score with %s threads
%s, retrying in %s seconds...
Evaluating %s Family Friendly Score with %s threads
%s, retrying in %s seconds...
Evaluating %s Family Friendly Score with %s threads
clean up pipeline due to error=DeploymentChecksError('None: None')
Shutdown handler de-registered
Shutdown handler registered
run pipeline %s
run pipeline stage %s
Running pipeline stage OfflineFamilyFriendlyScorer
Evaluating %s Family Friendly Score with %s threads
Pipeline stage OfflineFamilyFriendlyScorer completed in 3417.75s
Shutdown handler de-registered
chaiml-gemma3-27b-sft-exp1_v1 status is now inactive due to auto deactivation removed underperforming models
chaiml-gemma3-27b-sft-exp1_v1 status is now torndown due to DeploymentManager action