reef / config.json
seraphimserapis's picture
Initial commit
8e5599d
Raw History Blame Contribute Delete
7.57 kB
{
"name": "reef",
"format": "reef-v1",
"model": {
"state_len": 384,
"schema_len": 384,
"grid_width": 32,
"embed_dim": 256,
"input_dim": 128,
"width": 256,
"expansion": 3,
"kernel": [
3,
5
],
"depth": 11,
"dropout": 0.1,
"vocab_size": 29528,
"global_max": true,
"max_instruction_tokens": 128,
"max_option_tokens": 48,
"hoist_instructions": true,
"bands": 2,
"band_max": true,
"schema_to_state": true,
"mlp_rank": 0,
"hoist_empty_state": true,
"gate": true,
"zero_init": true,
"drop_opaque_keys": true,
"state_avgpool": true,
"mlp_leak": 0.1,
"mlp_norm": true,
"act_clip": {
"stem_q": [
-3.735,
4.0365
],
"polyp0_mixq": [
-33.7932,
38.4042
],
"polyp0_resq": [
-10.5252,
10.5416
],
"polyp1_mixq": [
-81.1884,
77.6882
],
"polyp1_resq": [
-19.6712,
16.6502
],
"polyp2_mixq": [
-198.8521,
196.2267
],
"polyp2_resq": [
-26.4767,
22.5516
],
"polyp3_mixq": [
-297.2089,
293.4478
],
"polyp3_resq": [
-35.1503,
34.8666
],
"polyp4_mixq": [
-396.9685,
418.4444
],
"polyp4_resq": [
-43.6214,
43.1923
],
"polyp5_mixq": [
-487.4022,
456.209
],
"polyp5_resq": [
-50.2699,
49.0352
],
"polyp6_mixq": [
-523.702,
544.392
],
"polyp6_resq": [
-54.3817,
54.8533
],
"polyp7_mixq": [
-565.9102,
583.926
],
"polyp7_resq": [
-58.6083,
59.2479
],
"polyp8_mixq": [
-595.5142,
627.5374
],
"polyp8_resq": [
-66.6176,
69.4399
],
"polyp9_mixq": [
-678.3539,
688.6672
],
"polyp9_resq": [
-74.3822,
76.3274
],
"polyp10_mixq": [
-772.6296,
768.8296
],
"polyp10_resq": [
-87.7964,
89.861
]
}
},
"temperatures": {
"noul": 1.3447,
"choice": 1.2085,
"score": 1.2748
},
"embeddings_source": "minishlab/potion-base-8M",
"encoder_params": 7379457,
"edgetpu": true,
"train_args": {
"train_data": [
"alert-triage",
"frigate-triage",
"notify-gate",
"protect-triage",
"unifi-network",
"open-jev:train?cap.painting-geometry-v1=4000&cap.snake-v1=3000&cap.vizdoom-basic-v1=2000&cap.tic_tac_toe-v1=1500&cap.tile_platformer-v1=1000",
"typed-decisions:train",
"train-v2",
"banking77_rest",
"civil_comments",
"clinc150_rest",
"contractnli",
"esci_rest",
"glaive_fc",
"go_emotions",
"home_appliance_pre",
"home_appliance_v2",
"massive_de",
"massive_en",
"minds14",
"mnli_oanc",
"sentfin_rest",
"sms_spam",
"snips",
"tfn_sentiment",
"tfn_topic",
"toolace_rest",
"wanli",
"when2call_rest",
"banking77",
"clinc150",
"when2call",
"apibank",
"bfcl_toolace",
"esci",
"isarcasm_a_en",
"sentfin",
"twitter_fin",
"hellaswag",
"winogrande",
"gsm8k"
],
"weights": {
"alert-triage": 2.0,
"frigate-triage": 2.0,
"notify-gate": 2.0,
"protect-triage": 2.0,
"unifi-network": 2.0,
"answer-audit": 2.0,
"agent_trace_observability": 4.0,
"customer_service": 4.0,
"invoice_processing": 4.0,
"security_incidents": 4.0,
"pretrain/": 0.12,
"pretrain/home_appliance_pre": 0.0,
"banking77": 0.5,
"clinc150": 0.5,
"esci": 0.5,
"when2call": 1.0,
"apibank": 0.0,
"bfcl_toolace": 1.0,
"home_appliance": 0.4,
"sentfin": 0.5,
"twitter_fin": 0.5,
"isarcasm": 0.5,
"hellaswag": 0.15,
"winogrande": 0.15,
"gsm8k": 0.15
},
"packing": {
"max_instruction_tokens": 128,
"max_option_tokens": 48,
"hoist_instructions": true,
"act_clip": {
"stem_q": [
-3.735,
4.0365
],
"polyp0_mixq": [
-33.7932,
38.4042
],
"polyp0_resq": [
-10.5252,
10.5416
],
"polyp1_mixq": [
-81.1884,
77.6882
],
"polyp1_resq": [
-19.6712,
16.6502
],
"polyp2_mixq": [
-198.8521,
196.2267
],
"polyp2_resq": [
-26.4767,
22.5516
],
"polyp3_mixq": [
-297.2089,
293.4478
],
"polyp3_resq": [
-35.1503,
34.8666
],
"polyp4_mixq": [
-396.9685,
418.4444
],
"polyp4_resq": [
-43.6214,
43.1923
],
"polyp5_mixq": [
-487.4022,
456.209
],
"polyp5_resq": [
-50.2699,
49.0352
],
"polyp6_mixq": [
-523.702,
544.392
],
"polyp6_resq": [
-54.3817,
54.8533
],
"polyp7_mixq": [
-565.9102,
583.926
],
"polyp7_resq": [
-58.6083,
59.2479
],
"polyp8_mixq": [
-595.5142,
627.5374
],
"polyp8_resq": [
-66.6176,
69.4399
],
"polyp9_mixq": [
-678.3539,
688.6672
],
"polyp9_resq": [
-74.3822,
76.3274
],
"polyp10_mixq": [
-772.6296,
768.8296
],
"polyp10_resq": [
-87.7964,
89.861
]
}
},
"select_on": [
"alert-triage",
"frigate-triage",
"notify-gate",
"protect-triage",
"unifi-network",
"answer-audit",
"agent_trace_observability",
"customer_service",
"invoice_processing",
"security_incidents",
"banking77",
"clinc150",
"esci",
"when2call",
"bfcl_toolace",
"apibank",
"home_appliance",
"sentfin",
"isarcasm"
],
"epochs": 1,
"lr": 0.0002,
"batch_size": 32,
"embeddings": "minishlab/potion-base-8M"
},
"float_temperatures": {
"noul": 1.2004,
"choice": 0.9816,
"score": 1.2004
},
"int8_calibration": {
"n_cases": 4000,
"n_questions": 13010,
"device": "cpu",
"select": [
"alert-triage",
"frigate-triage",
"notify-gate",
"protect-triage",
"unifi-network",
"answer-audit",
"agent_trace_observability",
"customer_service",
"invoice_processing",
"security_incidents",
"banking77",
"clinc150",
"esci",
"when2call",
"bfcl_toolace",
"apibank",
"home_appliance",
"sentfin",
"isarcasm"
],
"data": null,
"before": {
"n": 13010,
"accuracy": 0.5922367409684858,
"kl": 0.54532942520746,
"brier": 0.2630526837337706,
"ece": 0.04056949231805406
},
"after": {
"n": 13010,
"accuracy": 0.5922367409684858,
"kl": 0.5406201081100462,
"brier": 0.2634637770080148,
"ece": 0.06382528813215982
}
}
}