v0.10.0 (2024-07-15)

• Update manifest library and usage #275
• Support pointers when marshalling objects #277
• Add support for PostgreSQL host functions #278
• Fix dbpool reading after failed initialization #281
• Update for metadata changes #282
• Store function info with inference history #283
• Fix issues with GraphQL block quotes wundergraph/graphql-go-tools/843

v0.9.5 (2024-07-09)

• Use anonymous wasm modules for better performance #264
• Assume normalized vectors when calculating cosine similarity #265
• Support optional parameters #269
• Handle null parameters #270

v0.9.4 (2024-06-26)

• Increase batch size for auto-embedding collection texts #259
• Fix error with multiline input in GraphQL query #260

v0.9.3 (2024-06-25)

• Don't panic when the metadata DB is not configured #256
• Don't panic when collections are renamed or deleted #257

v0.9.2 (2024-06-24)

• Add auto-embedding for collection based on text checkpoint #250
• Remove extraneous types in graphql schemas #251
• Allow arrays as inputs to host functions #252
• Add batch upserts & batch recompute for collection & on auto-embedding #253

v0.9.1 (2024-06-22)

• Filter collection embedding functions from GraphQL schema #245
• Remove collection index from memory when manifest changes #246
• Fix missing execution id and plugin from logs from previous functions version #247
• Fix content type header when calling Hypermode-hosted models #248

v0.9.0 (2024-06-21)

• Add nullable check in ReadString #228
• Lowercase model name before invoking for hypermode hosted models #221
• Improve HTTP error messages #222
• Add host function for direct logging #224
• Refactoring, and add helpers for calling functions #226
• Add support for new model interface #229
• Add sequential vector search #240
• Update Hypermode-hosted model endpoint URL #242
• Fix bug caused by #226 #243

v0.8.2 (2024-06-23)

• Send backend ID with Sentry events #211 #213
• Add some logging for secrets #212
• Update logging to include Runtime version #215

v0.8.1 (2024-05-30)

• Fix compatibility with v1 authHeader secret #208
• Fix double-escaped JSON in OpenAI inference history #209

v0.8.0 (2024-05-29)

• Add Model Inference History to runtime #186
• Pass auth headers correctly when invoking a GraphQL API #196
• Use shared manifest module to read hypermode.json #199
• Pass HTTP auth secrets using v2 manifest format #203 #205