revert aml changes

This commit is contained in:
OlivierDehaene 2023-03-30 15:02:08 +02:00
parent 163c23f174
commit 3a0e706346

View File

@ -8,7 +8,7 @@ environment_variables:
MODEL_ID: bigscience/bloom
NUM_SHARD: 8
environment:
image: db4c2190dd824d1f950f5d1555fbadf0.azurecr.io/text-generation-inference:0.4.0
image: db4c2190dd824d1f950f5d1555fbadf0.azurecr.io/text-generation-inference:0.2.0
inference_config:
liveness_route:
port: 80
@ -18,21 +18,21 @@ environment:
path: /health
scoring_route:
port: 80
path: /
path: /generate
instance_type: Standard_ND96amsr_A100_v4
request_settings:
request_timeout_ms: 60000
request_timeout_ms: 90000
max_concurrent_requests_per_instance: 256
liveness_probe:
initial_delay: 90
timeout: 20
period: 60
initial_delay: 600
timeout: 90
period: 120
success_threshold: 1
failure_threshold: 2
failure_threshold: 5
readiness_probe:
initial_delay: 90
timeout: 60
period: 60
initial_delay: 600
timeout: 90
period: 120
success_threshold: 1
failure_threshold: 2
failure_threshold: 5
instance_count: 1