The daily firehose — everything the toolchain shipped today, already filtered.
// HOW THIS ISSUE IS MADE
We read every release from the 174 tools on our watchlist at the source — GitHub and GitLab release notes, vendor release pages and changelogs, project blogs and feeds, vendor press releases, and the source code behind the tag. Bug-fix-only releases and non-product newsroom noise are dropped; what's left is summarized down to the new capability, how to try it, and any screenshots or videos the release itself published. Every entry links to the sources it was built from.
Zed v0.96.2 adds PHP/Svelte/Bash LSP support, file icons, split resizing, and new editor settings.
└──▷ GET THIS VERSION
$ git clone --branch v0.96.2 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:$ git checkout v0.96.2
›Adds search::SelectAllMatches action (default Alt + Enter) to place carets and select all buffer search results simultaneously.
›Adds buffer_line_height setting to control editor line height.
›Adds indent_size setting to the project panel.
›Adds file and folder icons to the project panel, with an option to disable them via project panel settings.
›Adds split resizing to the center pane group; double-clicking a pane divider resets the split (disabled when active_pane_magnification is non-default).
+9 moreshow less
›Adds syntax highlighting and LSP support for PHP.
›Adds syntax highlighting and LSP support for Svelte.
›Adds syntax highlighting for Bash and shell scripts.
›Adds modifiers for opening files and symbols in a new pane, and for navigating to definitions and type definitions in a new pane.
›Adds rust-analyzer postfix completions and other completions with pre-resolved additional text edits.
›Adds a setting to mute the microphone on call join (enabled by default).
›Adds an option for configuring where the close button appears on editor tabs.
›Adds an option for showing git status on editor tabs.
›Adds ctrl+[ as a Vim-mode alias for Escape.
3 more releases in this issue
· 2023-07-05 → 2023-07-26
shell-gpt 0.9.4 lets you point --model at any OpenAI-compatible local API, enabling air-gapped LLM usage.
└──▷ GET THIS VERSION
$ git clone --branch 0.9.4 https://github.com/TheR1D/shell_gpt.git
# already have the repo? check out this version:$ git checkout 0.9.4
›Changes --model from an enum to a free-form string, allowing it to accept any model name including those served by self-hosted, OpenAI-compatible endpoints such as LocalAI.
›Supports locally hosted language models via any OpenAI-compatible API server (e.g. LocalAI), enabling offline or cost-free LLM usage on your own hardware.
└──▷ BREAKING ON UPGRADE
!The --model parameter type changed from an enum to a string; any tooling or scripts that relied on enum validation of --model values should be verified for compatibility.
AutoGPT v0.4.4 defaults to GPT-4, adds CLI args for agent identity, command aliases, and ships the long-awaited core re-arch.
└──▷ GET THIS VERSION
$ git clone --branch v0.4.4 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:$ git checkout v0.4.4
└──▷ TRY IT
Force GPT-3.5 across all tasks to control costs while experimenting with a new agent configuration.
$ python -m autogpt --gpt3only
Bootstrap a fully defined agent non-interactively in CI or scripted pipelines without an interactive prompt.
$ python -m autogpt --ai_name 'ReconBot' --ai_role 'OSINT researcher' --ai_goals 'Enumerate subdomains of example.com' 'Summarise findings to report.md'
›Adds --gpt3only and --gpt4only CLI flags to override model selection at runtime without changing config.
›Adds CLI args ai_name, ai_role, and ai_goals to define agent identity directly from the command line.
›SMART_LLM (formerly SMART_LLM_MODEL) now defaults to GPT-4 for high-accuracy tasks such as agent command selection; FAST_LLM (formerly FAST_LLM_MODEL) is used for lighter tasks like summarization.
›Introduces the autogpt/core module — the re-architecture project — now shipping in master and integrated starting with the Configuration subsystem.
›Adds command aliases to the agent command system, reducing the need for exact command-name matching.
+1 moreshow less
›Agent key bindings are now configurable via environment variables.
└──▷ BREAKING ON UPGRADE
!SMART_LLM now defaults to GPT-4 instead of the previous default, which will increase API costs for users who have not explicitly set a model.
›Adds farm-haystack[elasticsearch8] install extra and ElasticsearchDocumentStore auto-detection that selects the correct backend based on the installed Elasticsearch client version (covers ES 8 and ES <=7.5).
›Adds farm-haystack[elasticsearch7] install extra alongside the new elasticsearch8 extra for explicit version pinning.
›Introduces RecentnessRanker in haystack.nodes with date_meta_field, ranking_mode, and weight parameters to blend document age with relevance scores.
›Adds embed_meta_fields support to Ranker nodes, enabling metadata to be included in the text used for ranking.
›Adds support for list-typed embed_meta_fields when embedding metadata fields in retrievers.
+9 moreshow less
›Extends Anthropic Claude support to Claude 2 models with updated context window sizes and a new streaming API via PromptNode.
›Enables Llama 2 (including chat variant) on AWS SageMaker via PromptNode using aws_profile_name and aws_custom_attributes in model_kwargs.
›Upgrades dependency to transformers v4.31.0, enabling Llama 2 support for local inference.
›Adds global progress bar suppression capability to pipelines.
›Adds OpenAI-Organization header support for OpenAI authentication.
›Introduces LinkContentFetcher node by extracting link-retrieval logic from WebRetriever into a standalone component.
›Adds BM25 retrieval support for MemoryDocumentStore.
›Adds batch mode for MemoryRetriever (v2).
›Introduces a Store protocol (v2) and extends pipeline.add_component to support stores.
LangChain v0.0.247 adds Runnable.bind, RunnableMap, retry events, Few Shot Chat Prompt, and new LLM/embedding integrations.
└──▷ GET THIS VERSION
$ git clone --branch v0.0.247 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout v0.0.247
└──▷ USE IT
Attach a fixed stop sequence to any runnable so every invoke/stream/batch call uses it automatically.
python
from langchain.schema.runnable import RunnableLambda
base = RunnableLambda(lambda x: x)
bound = base.bind(stop=["\nObservation:"])
result = bound.invoke("What is 2+2?")
Build a parallel step with RunnableMap to fan out a single input to multiple runnables in one call.
python
from langchain.schema.runnable import RunnableMap, RunnableLambda
chain = RunnableMap({
"summary": RunnableLambda(lambda x: x["text"][:100]),
"length": RunnableLambda(lambda x: len(x["text"])),
})
result = chain.invoke({"text": "LangChain makes composing LLM pipelines easy."})
Use FewShotChatMessagePromptTemplate to inject labeled examples into a chat prompt before the user query.
$ git clone --branch v0.0.242 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout v0.0.242
›Adds AgentExecutorIterator to enable step-by-step iteration over agent execution, allowing callers to inspect or react to intermediate agent steps programmatically.
›Adds async support for TransformChain, enabling non-blocking use in async pipelines.
›Adds SelfQueryRetriever support for DeepLake vector store.
›Adds ArangoDB/AQL support to the Graph QA Chain via a new ArangoGraphQAChain integration.
›Adds EtherscanLoader document loader for pulling on-chain data into LangChain pipelines.
+7 moreshow less
›Adds LocalAIEmbeddings for generating embeddings via a locally hosted LocalAI instance.
›Adds a hybrid retriever that requires no external service, combining dense and sparse retrieval locally.
›Adds HuggingGPT integration for multi-model task orchestration via Hugging Face models.
›Adds stop sequence support to the Replicate LLM integration.
›Extends Cube Semantic Loader with additional functionality for richer semantic layer queries.
›Adds GPU and language setting controls to the NLP Cloud LLM integration.
›Adds filter parameter support to the Supabase vector store query, aligning with current Supabase API.
└──▷ BREAKING ON UPGRADE
!The default value of with_history for ChatGLM is changed to False.
LangChain v0.0.240rc1 adds kwargs support for Baseten models and sets up a new experimental module.
└──▷ GET THIS VERSION
$ git clone --branch v0.0.240rc1 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout v0.0.240rc1
›Adds kwargs support for Baseten models, enabling pass-through of additional parameters at invocation time.
›Sets up a new experimental package/module with its own release action, separating experimental features from the main library.
LangChain v0.0.240rc0 adds kwargs support for Baseten models and sets up an experimental module.
└──▷ GET THIS VERSION
$ git clone --branch v0.0.240rc0 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout v0.0.240rc0
›Adds kwargs support for Baseten models, enabling pass-through of arbitrary keyword arguments to the underlying model.
›Sets up a new experimental package/module, introducing a dedicated space for experimental LangChain features.
LangChain v0.0.226 adds HumanInputChatModel, Agent Trajectory evaluation, Load Evaluator, and a generic OpenAI function chain.
└──▷ GET THIS VERSION
$ git clone --branch v0.0.226 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout v0.0.226
└──▷ USE IT
Evaluate every step an agent took on a task, not just the final answer, using the new trajectory interface.
python
from langchain.evaluation import load_evaluator
evaluator = load_evaluator('trajectory')
result = evaluator.evaluate_agent_trajectory(
input='What is the capital of France?',
agent_trajectory=trajectory,
prediction=final_answer
)
print(result)
Limit how many DataFrame rows the pandas agent sees to reduce token usage on large datasets.
python
from langchain.agents import create_pandas_dataframe_agent
from langchain.llms import OpenAI
import pandas as pd
df = pd.read_csv('data.csv')
agent = create_pandas_dataframe_agent(OpenAI(temperature=0), df, number_of_head_rows=3)
agent.run('Which column has the most null values?')
Use HumanInputChatModel to manually drive a chain during local debugging without calling a live LLM.
python
from langchain.chat_models import HumanInputChatModel
from langchain.schema import HumanMessage
chat = HumanInputChatModel()
response = chat([HumanMessage(content='Summarize the risks in this contract.')])
print(response.content)
›Adds number_of_head_rows parameter to the pandas agent, letting callers control how many rows are shown to the agent for context.
›Adds HumanInputChatModel, a chat model implementation that accepts input from a human at the terminal — useful for testing and debugging chains interactively.
›Adds Agent Trajectory Interface for evaluating the full sequence of actions an agent takes, not just its final output.
›Adds Load Evaluator utility to instantiate evaluators by name at runtime without manually constructing them.
›Adds a generic OpenAI function chain, enabling structured function-calling workflows without writing a custom chain.
+7 moreshow less
›Adds elasticknn to the vector store init exports, making ElasticKNN available via the standard LangChain import path.
›Adds vector similarity search with scores to the Chroma vector store.
›Adds Re-use Trajectory Evaluator support, allowing a single trajectory evaluator instance to be applied across multiple runs.
›Adds automatic retry logic for Vertex LLM calls to handle transient API errors.
›Adds preset parameter to the TextGen LLM integration, allowing a named preset to be passed at invocation time.
›Enables PromptLayerChatOpenAI to support function call parameters, bringing it to parity with the base OpenAI chat model.
›Adds function call params to LLM invocation params so they are captured in run metadata and callbacks.
LangChain v0.0.221 adds Arthur, PromptLayer, and Flyte callback handlers, Zep auth, attachment support in UnstructuredEmailLoader, and a new Retriever interface with callbacks.
└──▷ GET THIS VERSION
$ git clone --branch v0.0.221 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout v0.0.221
›Enables UnstructuredEmailLoader to process email attachments, expanding document ingestion beyond the email body itself.
›Adds Arthur callback handler for tracking and monitoring LLM runs via the Arthur platform.
›Adds PromptLayer callback handler for logging and observability through PromptLayer.
›Adds Flyte callback handler for integrating LangChain runs into Flyte pipelines.
›Adds authentication support to the Zep memory integration.
+2 moreshow less
›Introduces a new Retriever interface with callback support, enabling observability hooks throughout retrieval.
›Adds parameter support on GoogleSearchApiWrapper for customizing search queries.
Retrieve per-generation performance details including stopping reason and token counts from the API.
$ curl http://localhost:5001/api/extra/perf
›Extends --hordeconfig to accept 5 parameters — [hordemodelname] [hordegenlength] [hordemaxctx] [hordeapikey] [hordeworkername] — which starts an embedded AI Horde worker that serves requests automatically in the background, eliminating the need for separate tools like HaidraScribe/KAIHordeBridge.
›Exposes --tensor_split for splitting model layers across multiple CUDA GPUs, matching upstream behavior.
›Retains --blasbatchsize 1024 support after it was removed upstream (note: scratch and KV buffer sizes will be larger when used).
›Adds LLAMA2 70B model support, with GQA automatically set to 8 upon detection.
›Adds additional fields to /api/extra/perf response, including stopping reason and generated token counts for the last generation.
+1 moreshow less
›Adds experimental Kepler architecture (e.g. K80) as a CUDA build target.
3 more releases in this issue
· 2023-07-07 → 2023-07-24
Query per-request token counter and performance metrics from the API for monitoring throughput.
$ curl http://localhost:5001/api/extra/perf
›Adds --ropeconfig <scale> <base> CLI argument to control both RoPE frequency scale (Linear) and RoPE frequency base (NTK-Aware) in a single flag, replacing the removed --linearrope flag — e.g. --ropeconfig 0.5 10000 for 2x linear scale or --ropeconfig 1.0 10000 for native LLAMA2 4K tuning.
›Exposes additional token counter data through the API endpoint /api/extra/perf.
›Automatically configures long-context NTK-Aware RoPE based on the --contextsize parameter by default, with no manual tuning required.
›Adds --ropeconfig support in the GUI for ease of use.
›Updates Kobold Lite with improved whitespace trim support and a new toggle for partial chat responses.
└──▷ BREAKING ON UPGRADE
!The --linearrope flag has been removed; replace it with --ropeconfig <scale> <base> (e.g. --ropeconfig 0.5 10000 for equivalent 2x linear scaling).
Poll real-time prompt-processing and generation timing from a running instance to benchmark throughput.
$ curl http://localhost:5001/api/extra/perf
›Adds --linearrope launch parameter to enable linear RoPE scaling (using 2048 as base) as an alternative to NTK-Aware RoPE; combine with --contextsize 8192 for a 0.25 linear scale suited to SuperHOT models.
›Exposes prompt-processing and generation timing via new API endpoint GET /api/extra/perf.
›Enables CUDA 8-bit MMV mode (quantized dot products) for formats q4_0, q4_1, q5_0, and q5_1, delivering significant GPU throughput gains when full GPU offload is used; K-quants and CL are unaffected.
›Adds Save and Load settings options to the GUI launcher.
›Adds 'All Devices' selection in the GUI for CUDA multi-GPU configurations.
+1 moreshow less
›Displays a warning when poor sampler orders are detected, nudging users toward the default configuration.
Use per-generation mirostat sampling via the /generate API instead of global defaults.
$ curl -X POST http://localhost:5001/api/v1/generate -H 'Content-Type: application/json' -d '{"prompt": "Once upon a time", "mirostat": 2, "mirostat_tau": 5.0, "mirostat_eta": 0.1, "sampler_order": [6,0,1,3,4,2,5]}'
›Adds --bantokens CLI flag to block a list of token substrings from being generated — e.g. --bantokens [ a ooo bans all tokens matching those substrings.
›Adds --usecublas lowvram <index> syntax for selecting a specific GPU by index when using CUDA multi-GPU setups.
›Adds sampler_order and mirostat/tau/eta parameters to the /generate API, settable per-generation request.
›Switches RoPE scaling to NTK-aware method driven by the existing --contextsize parameter, with support up to 8K context.
›Extends NTK-aware scaled RoPE support to GPT-NeoX and GPT-J model architectures, enabling longer context (e.g. 4K) on older models.
+3 moreshow less
›Adds 3K and 6K as selectable context size options alongside existing sizes.
›Adds a new customtkinter-based GUI with more configurable settings; requires the customtkinter Python module on Linux and macOS (bundled in Windows .exe builds).
›Kobold Lite now displays submitted contexts after each generation and adds two new scenarios plus limited Tavern v2 card support.
LocalAI v1.22.0 adds a new llama-master backend, JSONSchema ref resolution for planners, and a Llama 2 chat message template.
└──▷ GET THIS VERSION
$ git clone --branch v1.22.0 https://github.com/mudler/LocalAI.git
# already have the repo? check out this version:$ git checkout v1.22.0
›Adds the llama-master backend for running llama.cpp models via the latest upstream llama.cpp code.
›Adds a chat message template for Llama 2 models to correctly format multi-turn conversation prompts.
›Improves backend internals with general backends improvements in this release.
└──▷ BREAKING ON UPGRADE
!The backend formerly named llama-master is renamed to llama; the backend formerly named llama is renamed to llama-grammar — any model config or API calls referencing these backend names must be updated.
oobabooga textgen v1.4 adds llama-2-70b GGML support and expands OpenAI extension with images and logit_bias/logprobs.
└──▷ GET THIS VERSION
$ git clone --branch v1.4 https://github.com/oobabooga/textgen.git
# already have the repo? check out this version:$ git checkout v1.4
›OpenAI extension gains support for logit_bias and logprobs parameters, image handling, embeddings, and token endpoints — with improved error reporting and updated docs.
›Adds llama-2-70b GGML model support.
›Bumps exllama module to 0.0.8 with expanded LoRA support.
›Bumps bitsandbytes to 0.41.0 for faster inference speeds.
oobabooga textgen v1.3 adds Llama-v2 support, customizable RoPE for GGML, and LoRA loading via AutoGPTQ 0.3.0.
└──▷ GET THIS VERSION
$ git clone --branch v1.3 https://github.com/oobabooga/textgen.git
# already have the repo? check out this version:$ git checkout v1.3
›Adds Llama-v2 instruction template with auto-detection of truncation length.
›Adds customizable RoPE (Rotary Position Embedding) support for GGML models.
›Adds Airoboros-v1.2 instruction template.
›AutoGPTQ updated to 0.3.0, enabling LoRA loading out of the box for GPTQ models.
›LoRA menu selection is now preserved when loading a model (no longer reset on model load).
+1 moreshow less
›Disables 'Autoload the model' by default, and disables auto-loading at startup even when only one model is available.
└──▷ BREAKING ON UPGRADE
!The 'Autoload the model' setting is now disabled by default; setups that relied on automatic model loading at startup will need to re-enable it manually.
Triton v2.36.0 adds PyTorch implicit state, TF SavedModel direct serving, OpenTelemetry ensemble tracing, DLPack CUDA shared memory, Java bindings, and S3 repo scaling past 1000 files.
└──▷ GET THIS VERSION
$ git clone --branch v2.36.0 https://github.com/triton-inference-server/server.git
# already have the repo? check out this version:$ git checkout v2.36.0
›Adds model loading APIs to python_backend for BLS (Business Logic Scripting) usage.
›Supports direct serving of TensorFlow SavedModel via python_backend without a dedicated TF backend.
›Supports unpacked Conda execution environments in python_backend for custom dependency management.
›Adds DLPack tensor support to Triton Python client CUDA shared memory utilities.
›Adds implicit state management to pytorch_backend.
+3 moreshow less
›Extends OpenTelemetry trace mode to cover ensemble model tracing.
›Adds Java binding of the Triton in-process C++ API via javacpp-presets.
›Supports S3 model repositories containing more than 1000 files.
LanceDB v0.1.16 adds a Pydantic ORM layer with LanceModel and to_pydantic(), plus drop_table if-exists support.
└──▷ GET THIS VERSION
$ git clone --branch python-v0.1.16 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:$ git checkout python-v0.1.16
└──▷ USE IT
Define a typed vector schema with Pydantic and convert similarity-search results directly back to model instances.
›Adds LanceModel base class and vector() field type from lancedb.pydantic, enabling schema generation via LanceModel.to_arrow_schema() and round-tripping search results back to Pydantic models with .to_pydantic(<ModelClass>).
›Implements drop table if exists support.
›Makes pandas an optional dependency in LanceDB, reducing default install size.
8 more releases in this issue
· 2023-07-06 → 2023-07-31
LanceDB python-v0.1.12 passes the AWS_ENDPOINT environment variable for custom S3-compatible storage endpoints.
└──▷ GET THIS VERSION
$ git clone --branch python-v0.1.12 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:$ git checkout python-v0.1.12
›Supports the AWS_ENDPOINT environment variable to direct LanceDB at custom S3-compatible storage backends (e.g. MinIO, LocalStack).
$ git clone --branch v0.1.11-python https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:$ git checkout v0.1.11-python
›Supports creating a table by passing an Iterator[RecordBatch] as the data source, enabling streaming ingestion of large datasets.
›Adds conversion of Pydantic models to Arrow Schema, letting callers define table structure with typed Python models.
›Adds schema serialization to JSON via a new schema-to-JSON conversion path.
›Exposes table schema and version in the Rust layer, surfacing them through the Python get table schema API.
›Enables listing tables from a remote LanceDB service via the Python client.
+1 moreshow less
›Supports adding records to a remote table via the Python remote API.
LanceDB v0.1.10 adds empty table creation and changes the default write mode to error on conflict.
└──▷ GET THIS VERSION
$ git clone --branch v0.1.10-python https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:$ git checkout v0.1.10-python
›Changes the default write mode from drop to error, so accidental overwrites now raise an error instead of silently dropping data.
›Supports creation of an empty table without requiring initial data to be provided.
›AWS credentials are now cached until 30 seconds before expiry, reducing redundant credential fetches in cloud-backed datasets.
└──▷ BREAKING ON UPGRADE
!The default write mode is changed from drop to error: existing code that relied on the silent drop-and-overwrite behavior will now raise an error on conflicting writes.
›Exposes IVF PQ index configuration in the Node.js client, letting callers tune partitioning and quantization parameters when building vector indexes.
›Adds replace flag to the JavaScript createIndex API, allowing an existing index to be overwritten in place without dropping the table.
›Supports WriteMode in the Node.js createTable API (re-exported from lancedb in Rust), enabling append, overwrite, or create-or-append semantics at table creation time.
›Supports specifying a named vector column for vector search, so tables with multiple vector columns can target the correct one explicitly.
›Adds dot product distance metric support in the JavaScript/Node.js client for vector similarity search.
+1 moreshow less
›Makes the object store construction hook public, enabling custom storage backend injection.
Milvus 2.2.12 adds high-level RESTful APIs, vector retrieval by ID, json_contains filtering, and GCS/OSS access-key support.
└──▷ GET THIS VERSION
$ git clone --branch v2.2.12 https://github.com/milvus-io/milvus.git
# already have the repo? check out this version:$ git checkout v2.2.12
›Adds a high-level RESTful API that listens on the same port as gRPC, simplifying client-side operations (note: a token must be set even when authentication is disabled).
›Adds json_contains expression support for filtering on JSON fields in searches and queries.
›Enables bulk-insert to support partition keys.
›Enables the chunk manager to use GCS and OSS object storage with an access key.
›Adds minCPUParallelTaskNumRatio config to improve parallelism when a single task's estimated CPU usage exceeds total CPU capacity.
+5 moreshow less
›Supports setting the vector field as an output field in ANN searches and queries against HNSW-, DiskANN-, or IVF-FLAT-indexed collections.
›Makes compaction RPC timeout and maximum parallelism configurable.
›Writes cache files to the cacheStorage.rootpath directory.
›Adds a PK index for string data types, improving query performance on string primary keys.
›Introduces native multi-tenancy with strong tenant isolation, supporting 50,000+ tenants per node and millions of tenants with billions of objects in a multi-node cluster; enable via class schema configuration.
›Adds GET /tenants endpoint to list tenants of a multi-tenant class, plus endpoints to create and delete tenants for a specific class.
›Supports full single-tenant object CRUD, batch operations, and batch reference operations, with tenant key immutability enforced.
›Extends the nodes API to surface multi-tenant class information.
›Adds multi-tenancy support to GQL Get{} and GQL Aggregate{} queries, including nearObject and nearText with tenant context.
+8 moreshow less
›Adds replication support for multi-tenant classes.
›Enables Prometheus metrics for classes with multi-tenancy enabled.
›Introduces autocut for bm25, nearVector, nearObject, and nearXXX queries to automatically cut off unrelated results.
›Adds autocut and a RelativeScore fusion algorithm to hybrid search for improved result quality.
›Introduces reranker-transformers module for post-retrieval reranking using transformer models.
›Introduces reranker-cohere module for post-retrieval reranking using the Cohere API.
›Adds status code metrics distinguishing OK, user error, and server error responses for better observability of request success and failure rates.
›Product Quantization (PQ) moves to general availability, with dynamic rescoring of results and a configurable training limit.
THE AI TOOLCHAIN