One of up to four strings that make the model halt generation; the sequence itself is not included in the output.
Also called stop.
Read more: Microsoft Learn
In the Ultra Transcenders books
Each book explains Stop sequence in context, with comparison tables and the common traps.
Terms in this definition
- SEQUENCE
A schema-bound object that generates values on request via NEXT VALUE FOR without belonging to a table. One can feed several tables, restart, cycle or set aside a block of values; any value drawn inside a transaction that rolls back is gone.
Related terms
- 168.63.129.16
Special platform address (WireServer) through which Azure delivers the VM agent channel, built-in DNS, DHCP and load balancer health probes. Traffic to it must stay open, or those platform functions stop working.
- Access keys
Pair of 512-bit keys belonging to a storage account; every account SAS and service SAS is signed with one of them. Once both are regenerated, direct key access and all those SAS tokens stop working.
- Agent risk
Condition in Conditional Access based on the risk level Microsoft Entra ID Protection assigns to agent identities, letting a policy stop high-risk agents from obtaining tokens.
- Attack surface reduction
Protections in Defender for Endpoint that close off behaviour attackers like to abuse. Chief among them are ASR rules, which on Windows can stop Office programs spawning child processes or obfuscated scripts from running.
- Automatic assignment policy
Works much like a dynamic group: a membership rule based on user attributes decides who gets an access package and removes assignments when people stop matching. It is an ID Governance entitlement management feature, and neither catalog owners nor access package managers are allowed to set one up.
- Blocklist
A content filter can reference this user-defined collection of terms or regular expressions, holding as many as 10,000 entries, to stop particular words getting through.
- Branch protection
Rules on a branch such as main (branch protection in GitHub, branch policies in Azure Repos) that stop people pushing to it directly and require reviewed pull requests for merges.
- Capacity pools
Custom Spark pools defined by a capacity admin in the Spark Compute settings of a capacity, which then show up as choices for workspaces and environments and take roughly two to three minutes to start. From the same place the admin can stop workspaces customising pools or using the starter pool.