CHANGELOG.md

Changelog

4.56

• [#400] (cohere-ai#400)
  • Remove unsupported chat parameters

4.55

• [#401] (cohere-ai#401)
  • Update to API temperature defaults in chat

4.54

• [#389] (cohere-ai#397)
  • Add raw_prompting param to chat

4.53

• [#389] (cohere-ai#392)
  • Add tools and tool_results to chat

4.52

• [#389] (cohere-ai#389)
  • Rename preamble_override to preamble in chat parameters

4.51

• [#388] (cohere-ai#388)
  • Increase timeout to 300s

4.50

• [#384] (cohere-ai#386)
  • Remove deprecated Detect Language API

4.49

• [#384] (cohere-ai#384)
  • Add finish_reason to non stream chat

4.48

• [#380] (cohere-ai#380)
  • Remove logit_bias parameter from Generate

4.47

• [#378] (cohere-ai#378)
  • Adds embedding_types to embed job

4.46

• [#376] (cohere-ai#376)
  • Adds avro to dataset save formats

4.45

• [#372] (cohere-ai#372)
  • Remove logit_bias parameter from Chat

4.44

• [#366] (cohere-ai#366)
  • Update embed job list request

4.43

• [#365] (cohere-ai#365)
  • Update embed job parameters

4.42

• [#359] (cohere-ai#359)
  • Add embedding_types to embed request
  • Update embed response with embeddings_by_types response

4.41

• [#358] (cohere-ai#363)
  • Update dataset route to datasets

4.40

• [#357] (cohere-ai#361)
  • Add iter to dataset

4.39

• [#357] (cohere-ai#357)
  • Embed: add bulk embed support for embed V3

4.38

• [#351] (cohere-ai#351)
  • Add connector support
• [#356] (cohere-ai#356)
  • Add connector oauth authorize support
• [#345] (cohere-ai#345)
  • Add mapping for status events for finetunes

4.37

• [#342] (cohere-ai#342)
  • Update Poetry, remove Python 3.7 support

4.36

• [#309] (cohere-ai#309)
  • Dataset: use dataset api in creating fine-tuned model

4.35

• [#339] (cohere-ai#339)
  • Dataset: default to train epochs

4.34

• [#338] (cohere-ai#338)
  • Dataset: add get usage

4.33

• [#337] (cohere-ai#337)
  • Dataset: update data param

4.32

• [#331] (cohere-ai#331)
  • Embed: add compression parameter for embed models

4.31

• [#324] (cohere-ai#324)
  • Classify:
    • Deprecate prediction and confidence attribute
    • Add new predictions and confidences attribute for single and multi label classification

4.30

• [#313] (cohere-ai#313)
  • change chatlog (string) to chat_history (array of messages) in /chat
• [#312] (cohere-ai#312)
  • add prompt_truncation in chat tests
• [#321] (cohere-ai#321)
  • remove generate finish reason test assertion
• [#322] (cohere-ai#322)
  • remove unneeded max_tokens test case

4.29

• [#311] (cohere-ai#311)
  • Embed: remove embed input_type tests

4.28

• [#310] (cohere-ai#310)
  • Embed: add input_type parameter for new embed models

4.27

• [#308] (cohere-ai#308)
  • Datasets: add validation_warnings

4.26.1

• [#306] (cohere-ai#306)
  • AsyncClient: Fix correctly raising error on connection issues.

4.26

• [#296] (cohere-ai#301)
  • Chat: add support for prompt_truncation param

4.25

• [#303] (cohere-ai#303)
  • Allow uploading of evaluation data

4.24

• [#296] (cohere-ai#296)
  • Allow passing of delimiter for csv

4.23

• [#294] (cohere-ai#294)
  • Allow passing of ParseInfo for datasets

4.22

• [#292] (cohere-ai#292)
  • Add search query only parameter
  • Add documents to generate citations
  • Add connectors to generate citations
  • Add citation quality

4.21

• [#287] (cohere-ai#287)
  • Remove deprecated chat "query" parameter including inside chat_history parameter
  • Support event-type for chat streaming

4.20.2

• [#284] (cohere-ai#284)
  • Rename dataset urls to download_urls

4.20.1

• [#279] (cohere-ai#279)
  • Fix dataset listing key error

4.20.0

• [#276] (cohere-ai#276)
  • Add support for base_model option in create_custom_model
• [#277] (cohere-ai#277)
  • Add support for co.loglikelihood endpoint

4.19.1

• [#273] (cohere-ai#273)
  • Fix fastavro version for python >=3.8

4.19

• #255
  • Add custom model metrics endpoint

4.18

• #268
  • Add dataset endpoint

4.17

• #263
  • Add wait() to custom models

4.16

• #262
  • Deprecate embed custom models

4.15

• #256
  • Add rerank finetuning endpoint

4.14

• #259
  • Add the api_url parameter to the client.

4.13

• #258
  • Add generate_descriptions parameter to create_cluster_job
• #261
  • Pin urllib3 version

4.12

• #248
  • Add deprecation warning for Chat parameter query
  • Add deprecation warning for Chat text in chat_history
  • Deprecate Chat chatlog_override
• #254
  • Add support for hyperparameters for custom models

4.11

• #245
  • Add parameters p, k and logit_bias to chat
• #228
  • Better string representation for DetectLanguageResponse
• #249
  • Catch ClientPayloadError in AsyncClient and convert it to a CohereAPIError

4.10.0

• #247
  • Remove id from intermediate streaming response (StreamingText) from generate
  • Add is_finished to intermediate streaming response (StreamingText) from generate
  • Add id to StreamingGenerations final response from generate
  • Add finish_reason to StreamingGenerations final response from generate
  • Add Generations to StreamingGenerations final response from generate

4.9.1

• #222
  • Pass along SDK version in request source header

4.9.0

• #242
  • Add token count to the chat response

4.8.0

• #237
  • Add support for custom models

4.7.0

• #232
  • Support model parameter in tokenize and detokenize

4.6.1

• #240
  • Revert: Add SDK level validation for classify params

4.6.0

• #238
  • Add is_finished to each element of the streaming chat response
  • Add conversation_id, response_id, finish_reason, chatlog, preamble and prompt to the streaming chat response
  • Fix chat streaming index

4.5.1

• #225
  • Remove support for the co.chat parameter chatlog_override and add deprecation warning

4.5.0

• #229
  • Add return_exceptions parameter to Client's batch_* methods, mirroring AsyncClient

4.4.2

• #230
  • Add SDK level validation for classify params

4.4.1

• #224
  • Update co.chat parameter chat_history

4.4.0

• #223
  • Remove deprecated co.chat parameter reply

4.3.1

• #220
  • Update chat params
    • Add support for chat_history

4.3.0

• #210
  • Update embed with compressed embeddings
    • compress
    • compression_codebook
• #211
  • Add co.codebook endpoint for compressed embeddings

4.2.1

• #214
  • Add support for co.Chat parameter:
    • return_preamble

4.2.0

• #212
  • Deprecate co.chat params
    • session_id
    • persona_name
    • persona_prompt
  • Add deprecation warning for Chat attribute
    • Use text instead of reply
  • Add support for generation_id
• #206
  • Update cluster endpoint to use UMAP+HDBSCAN
  • Remove threshold and add n_neighbors and is_deterministic as params

4.1.6

• #205
  • Add param max_chunks_per_doc to rerank
  • Enforce model param for rerank

4.1.5

• #208
  • Fix a missing import for CohereConnectionError

4.1.4

• #204
  • Add generate_preference_feedback for submitting preference-style feedback

4.1.3

• #194
  • Return the generation ID for chat

4.1.2

• #192
  • Fix duplicate Generate calls in the sync SDK

4.1.1

• #190
  • Remove wrong "Embedding" class used for type hinting

4.1.0

• #188
  • Add stream parameter to chat, and relevant return object.
• #169
  • Add stream parameter to generate, and relevant return object.
  • Add example notebook for streaming.

4.0.6

• #187
  • Refactor feedback to be generate specific

4.0.5

• #186
  • Added warnings support for meta response

4.0.4

• #185
  • Validate API key without API call

4.0.3

• #184
  • Respect timeout option for sync client

4.0.2

• #183
  • Better error messages for synchronous client

4.0.1

• #181
  • Allow Python >=3.11

4.0

• #160
  • Add AsyncClient
  • Default value of API key from environment variable CO_API_KEY.
  • Feedback endpoint moved from CohereObject to Client/AsyncClient.
  • Lazy initialization using futures removed.
  • Generations is now a UserList, and initialized from responses using from_dict.
  • Chat objects are initialized using from_dict. Optional attributes are now None rather than missing.
  • Documentation expanded and built using sphinx.
  • Use Poetry, and format using black and isort, include pre-commit hooks.
  • Removed ability for user to choose API version. This SDK version defaults to v1.
  • Added 'meta' fields to response objects with API version
• #179
  • Add support for co.Chat parameters: temperature, max_tokens, persona_name, persona_prompt
  • Remove support for co.Chat parameters: persona, preamble_override
  • Updates the co.Chat user_name parameter

3.10.0

• #176
  • Add failure reason to clustering jobs

3.9.1

• #175
  • Fix url path for cluster-job get endpoint

3.9

• #168
  • Add support for co.Rerank parameter:
    • model

3.8

• #158
  • Add support for co.Chat parameters:
    • preamble_override
    • return_prompt
    • username

3.7

• #164
  • Add clustering methods:
    • co.create_cluster_job
    • co.get_cluster_job
    • co.list_cluster_jobs
    • co.wait_for_cluster_job

3.6

• #156
  • Replace abstractiveness param with extractiveness in co.Summarize
  • Rename additional_instruction param to additional_command in co.Summarize

3.5

• #157
  • Add support for chatlog_override parameter for co.Chat

3.4

• #154
  • Add support for return_chatlog parameter for co.Chat

3.3

• #146
  • Add new experimental co.rerank api
• #150
  • Add abstractiveness param for co.summarize

3.2

• #138
  • Add model to the chat endpoint's parameters
• #145
  • Add new experimental co.summarize API

3.1

• #129
  • Add support for end_sequences param in Generate API

3.1

• #126
  • Add new co.detect_language api

3.0

• #125
  • Improve the Classify response string representation

2.9

• #120
  • Remove experimental Extract API from the SDK

2.8

• #112
  • Add support for prompt_vars parameter for co.Generate

2.7

• #110
  • Classification.confidence is now a float instead of a list

2.6

• #105
  • Remove deprecated options from classify
• #104
  • Remove experimental moderate api

2.5

• #96
  • The default max_tokens value is now configured on the backend

2.4

• #102
  • Generate Parameter now accepts logit_bias as a parameter

2.2

• #95
  • Introduce Detokenize for converting a list of tokens to a string
• #92
  • Handle truncate parameter for Classify and Generate

1.3

• #71 Sunset Choose Best

1.0

• #38
  • Handle truncate parameter for Embed
• #36 Change generations to return Generations, which has as a list of Generation * Each Generation has a text and token_likelihoods field to store generations and token likelihoods respectively
• #34 API Updates and SDK QoL Improvements _Add support for multiple generations _ Add capability to use a specific API version * Fully remove CohereClient
• #32 Handle different errors more safely

0.0.13 - 2021-08-31

• #26 Add Request Source

0.0.11 - 2021-07-22

• #24 SDK QoL Updates
  • Change from CohereClient to be Client –– the CohereClient will be completely deprecated in the future
  • Have a more human-friendly output when printing Cohere response objects directly

0.0.10 - 2021-07-20

• #23 Add token_log_likelihoods to the Choose Best endpoint
• #21 Change from BestChoices.likelihoods to BestChoices.scores
• #19 Make some response objects be iterable

0.0.9 - 2021-06-14

• #18 API Updates - Generate Endpoint
  • Add Frequency Penalty, Presence Penalty, Stop Sequences, and Return Likelihoods for Generate