Skip to content

Commit 760bc6c

Browse files
cpsievertcpsievert
committed
Merge branch 'main' into vllm
2 parents 5033d06 + 0b2ae52 commit 760bc6c

File tree

204 files changed

+17622
-3370
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

204 files changed

+17622
-3370
lines changed

‎.github/workflows/check-update-types.yml‎

Lines changed: 6 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
name: Test - Python
1+
name: Check Provider Types
22

33
on:
44
workflow_dispatch:
@@ -36,7 +36,12 @@ jobs:
3636
# Fail if there are changes under chatlas/types and throw a useful error message
3737
- name: Check for changes
3838
run: |
39+
echo "Checking for changes in types..."
40+
git status chatlas/types
41+
git diff --stat chatlas/types
3942
if [[ -n $(git diff chatlas/types) ]]; then
43+
echo "Changes detected:"
44+
git diff chatlas/types
4045
echo "::error::Types are out of date. Please run 'make update-types' and commit the changes."
4146
exit 1
4247
fi

‎.github/workflows/release.yml‎

Lines changed: 1 addition & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -5,8 +5,7 @@ on:
55
types: [published]
66

77
env:
8-
UV_VERSION: "0.4.x"
9-
PYTHON_VERSION: 3.13
8+
PYTHON_VERSION: 3.12
109

1110
jobs:
1211
pypi-release:
@@ -27,8 +26,6 @@ jobs:
2726

2827
- name: 🚀 Install uv
2928
uses: astral-sh/setup-uv@v3
30-
with:
31-
version: ${{ env.UV_VERSION }}
3229

3330
- name: 🐍 Set up Python ${{ env.PYTHON_VERSION }}
3431
run: uv python install ${{ env.PYTHON_VERSION }}

‎.github/workflows/test.yml‎

Lines changed: 21 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -29,6 +29,8 @@ jobs:
2929
ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
3030
GOOGLE_API_KEY: ${{ secrets.GOOGLE_API_KEY }}
3131
AZURE_OPENAI_API_KEY: ${{ secrets.AZURE_OPENAI_API_KEY }}
32+
DATABRICKS_HOST: ${{ secrets.DATABRICKS_HOST }}
33+
DATABRICKS_TOKEN: ${{ secrets.DATABRICKS_TOKEN }}
3234
# Free tier of Google is rate limited, so we only test on 3.12
3335
TEST_GOOGLE: ${{ matrix.config.test_google }}
3436
# Free tier of Azure is rate limited, so we only test on 3.12
@@ -46,10 +48,29 @@ jobs:
4648
- name: 📦 Install the project
4749
run: uv sync --python ${{ matrix.config.python-version }} --all-extras
4850

51+
- name: Create Snowflake connections.toml file
52+
run: |
53+
mkdir -p ~/.ssh
54+
echo '${{ secrets.SNOWFLAKE_SSH_PUBLIC_KEY }}' > ~/.ssh/snowflake_key.pub
55+
echo '${{ secrets.SNOWFLAKE_SSH_PRIVATE_KEY }}' > ~/.ssh/snowflake_key.p8
56+
chmod 644 ~/.ssh/snowflake_key.pub
57+
chmod 600 ~/.ssh/snowflake_key.p8
58+
chmod 700 ~/.ssh
59+
60+
mkdir -p ~/.snowflake
61+
cat > ~/.snowflake/connections.toml << 'EOF'
62+
[posit]
63+
account="duloftf-posit-software-pbc-dev"
64+
user="[email protected]"
65+
private_key_file="/home/runner/.ssh/snowflake_key.p8"
66+
private_key_file_pwd="${{ secrets.SNOWFLAKE_SSH_PRIVATE_KEY_PASSWORD }}"
67+
EOF
68+
4969
- name: 🧪 Check tests
5070
run: make check-tests
5171

5272
- name: 📝 Check types
73+
if: ${{ matrix.config.python-version != '3.9' }}
5374
run: make check-types
5475

5576
- name: 📐 Check formatting

‎.github/workflows/update-pricing.yml‎

Lines changed: 38 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,38 @@
1+
name: Update Pricing
2+
3+
on:
4+
workflow_dispatch:
5+
pull_request:
6+
types: [opened, synchronize, reopened, ready_for_review]
7+
8+
jobs:
9+
check-pricing:
10+
name: Check for pricing updates in Ellmer
11+
runs-on: ubuntu-latest
12+
13+
steps:
14+
15+
- name: Checkout current prices.json in chatlas
16+
uses: actions/checkout@v4
17+
with:
18+
sparse-checkout: /chatlas/data/prices.json
19+
sparse-checkout-cone-mode: false
20+
path: main
21+
22+
- name: Get Ellmer prices.json
23+
uses: actions/checkout@v4
24+
with:
25+
sparse-checkout: /data-raw/prices.json
26+
sparse-checkout-cone-mode: false
27+
repository: tidyverse/ellmer
28+
path: ellmer
29+
30+
- name: Check for differences
31+
run: |
32+
echo "Checking diff between prices.json"
33+
git diff --no-index --stat ellmer/data-raw/prices.json main/chatlas/data/prices.json
34+
if [[ -n $(git diff --no-index --stat ellmer/data-raw/prices.json main/chatlas/data/prices.json) ]]; then
35+
echo "Changes detected:"
36+
echo "::error::Ellmer's prices.json does not match the current Chatlas prices.json"
37+
exit 1
38+
fi

‎.gitignore‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -10,5 +10,5 @@ sandbox/
1010

1111
/.luarc.json
1212

13-
# Automatically generated by docs/index.py
14-
docs/index.qmd
13+
# setuptools_scm
14+
chatlas/_version.py

‎CHANGELOG.md‎

Lines changed: 195 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -11,14 +11,207 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
1111

1212
### New features
1313

14-
* Adds vLLM support via a new `ChatVLLM` class. (#24)
14+
* Added `ChatDeepSeek()` for chatting via [DeepSeek](https://www.deepseek.com/). (#147)
15+
* Added `ChatOpenRouter()` for chatting via [Open Router](https://openrouter.ai/). (#148)
16+
* Added `ChatHuggingFace()` for chatting via [Hugging Face](https://huggingface.co/). (#144)
17+
* Added `ChatPortkey()` for chatting via [Portkey AI](https://portkey.ai/). (#143)
18+
* Added `ChatVllm()` for chatting via [vLLM](https://docs.vllm.ai/en/latest/). (#24)
19+
20+
### Bug fixes
21+
22+
* Fixed an issue where chatting with some models was leading to `KeyError: 'cached_input'`. (#149)
23+
24+
## [0.9.2] - 2025-08-08
25+
26+
### Improvements
27+
28+
* `Chat.get_cost()` now covers many more models and also takes cached tokens into account. (#133)
29+
* Avoid erroring when tool calls occur with recent versions of `openai` (> v1.99.5). (#141)
30+
31+
32+
## [0.9.1] - 2025-07-09
33+
34+
### Bug fixes
35+
36+
* Fixed an issue where `.chat()` wasn't streaming output properly in (the latest build of) Positron's Jupyter notebook. (#131)
37+
38+
* Needless warnings and errors are no longer thrown when model pricing info is unavailable. (#132)
39+
40+
## [0.9.0] - 2025-07-02
41+
42+
### New features
43+
44+
* `Chat` gains a handful of new methods:
45+
* `.register_mcp_tools_http_stream_async()` and `.register_mcp_tools_stdio_async()`: for registering tools from a [MCP server](https://modelcontextprotocol.io/). (#39)
46+
* `.get_tools()` and `.set_tools()`: for fine-grained control over registered tools. (#39)
47+
* `.set_model_params()`: for setting common LLM parameters in a model-agnostic fashion. (#127)
48+
* `.get_cost()`: to get the estimated cost of the chat. Only popular models are supported, but you can also supply your own token prices. (#106)
49+
* `.add_turn()`: to add `Turn`(s) to the current chat history. (#126)
50+
* Tool functions passed to `.register_tool()` can now `yield` numerous results. (#39)
51+
* A `ContentToolResultImage` content class was added for returning images from tools. It is currently only works with `ChatAnthropic`. (#39)
52+
* A `Tool` can now be constructed from a pre-existing tool schema (via a new `__init__` method). (#39)
53+
* The `Chat.app()` method gains a `host` parameter. (#122)
54+
* `ChatGithub()` now supports the more standard `GITHUB_TOKEN` environment variable for storing the API key. (#123)
55+
56+
### Changes
57+
58+
#### Breaking Changes
59+
60+
* `Chat` constructors (`ChatOpenAI()`, `ChatAnthropic()`, etc) no longer have a `turns` keyword parameter. Use the `.set_turns()` method instead to set the (initial) chat history. (#126)
61+
* `Chat`'s `.tokens()` methods have been removed in favor of `.get_tokens()` which returns both cumulative tokens in the turn and discrete tokens. (#106)
62+
63+
#### Other Changes
64+
65+
* `Tool`'s constructor no longer takes a function as input. Use the new `.from_func()` method instead to create a `Tool` from a function. (#39)
66+
* `.register_tool()` now throws an exception when the tool has the same name as an already registered tool. Set the new `force` parameter to `True` to force the registration. (#39)
67+
68+
### Improvements
69+
70+
* `ChatGoogle()` and `ChatVertex()` now default to Gemini 2.5 (instead of 2.0). (#125)
71+
* `ChatOpenAI()` and `ChatGithub()` now default to GPT 4.1 (instead of 4o). (#115)
72+
* `ChatAnthropic()` now supports `content_image_url()`. (#112)
73+
* HTML styling improvements for `ContentToolResult` and `ContentToolRequest`. (#39)
74+
* `Chat`'s representation now includes cost information if it can be calculated. (#106)
75+
* `token_usage()` includes cost if it can be calculated. (#106)
76+
77+
### Bug fixes
78+
79+
* Fixed an issue where `httpx` client customization (e.g., `ChatOpenAI(kwargs = {"http_client": httpx.Client()})`) wasn't working as expected (#108)
80+
81+
### Developer APIs
82+
83+
* The base `Provider` class now includes a `name` and `model` property. In order for them to work properly, provider implementations should pass a `name` and `model` along to the `__init__()` method. (#106)
84+
* `Provider` implementations must implement two new abstract methods: `translate_model_params()` and `supported_model_params()`.
85+
86+
## [0.8.1] - 2025-05-30
87+
88+
* Fixed `@overload` definitions for `.stream()` and `.stream_async()`.
89+
90+
## [0.8.0] - 2025-05-30
91+
92+
### New features
93+
94+
* New `.on_tool_request()` and `.on_tool_result()` methods register callbacks that fire when a tool is requested or produces a result. These callbacks can be used to implement custom logging or other actions when tools are called, without modifying the tool function (#101).
95+
* New `ToolRejectError` exception can be thrown from tool request/result callbacks or from within a tool function itself to prevent the tool from executing. Moreover, this exception will provide some context for the the LLM to know that the tool didn't produce a result because it was rejected. (#101)
96+
97+
### Improvements
98+
99+
* The `CHATLAS_LOG` environment variable now enables logs for the relevant model provider. It now also supports a level of `debug` in addition to `info`. (#97)
100+
* `ChatSnowflake()` now supports tool calling. (#98)
101+
* `Chat` instances can now be deep copied, which is useful for forking the chat session. (#96)
102+
103+
### Changes
104+
105+
* `ChatDatabricks()`'s `model` now defaults to `databricks-claude-3-7-sonnet` instead of `databricks-dbrx-instruct`. (#95)
106+
* `ChatSnowflake()`'s `model` now defaults to `claude-3-7-sonnet` instead of `llama3.1-70b`. (#98)
107+
108+
### Bug fixes
109+
110+
* Fixed an issue where `ChatDatabricks()` with an Anthropic `model` wasn't handling empty-string responses gracefully. (#95)
111+
112+
113+
## [0.7.1] - 2025-05-10
114+
115+
* Added `openai` as a hard dependency, making installation easier for a wide range of use cases. (#91)
116+
117+
## [0.7.0] - 2025-04-22
118+
119+
### New features
120+
121+
* Added `ChatDatabricks()`, for chatting with Databrick's [foundation models](https://docs.databricks.com/aws/en/machine-learning/model-serving/score-foundation-models). (#82)
122+
* `.stream()` and `.stream_async()` gain a `content` argument. Set this to `"all"` to include `ContentToolResult`/`ContentToolRequest` objects in the stream. (#75)
123+
* `ContentToolResult`/`ContentToolRequest` are now exported to `chatlas` namespace. (#75)
124+
* `ContentToolResult`/`ContentToolRequest` gain a `.tagify()` method so they render sensibly in a Shiny app. (#75)
125+
* A tool can now return a `ContentToolResult`. This is useful for:
126+
* Specifying the format used for sending the tool result to the chat model (`model_format`). (#87)
127+
* Custom rendering of the tool result (by overriding relevant methods in a subclass). (#75)
128+
* `Chat` gains a new `.current_display` property. When a `.chat()` or `.stream()` is currently active, this property returns an object with a `.echo()` method (to echo new content to the display). This is primarily useful for displaying custom content during a tool call. (#79)
129+
130+
### Improvements
131+
132+
* When a tool call ends in failure, a warning is now raised and the stacktrace is printed. (#79)
133+
* Several improvements to `ChatSnowflake()`:
134+
* `.extract_data()` is now supported.
135+
* `async` methods are now supported. (#81)
136+
* Fixed an issue with more than one session being active at once. (#83)
137+
* `ChatAnthropic()` no longer chokes after receiving an output that consists only of whitespace. (#86)
138+
* `orjson` is now used for JSON loading and dumping. (#87)
139+
140+
### Changes
141+
142+
* The `echo` argument of the `.chat()` method defaults to a new value of `"output"`. As a result, tool requests and results are now echoed by default. To revert to the previous behavior, set `echo="text"`. (#78)
143+
* Tool results are now dumped to JSON by default before being sent to the model. To revert to the previous behavior, have the tool return a `ContentToolResult` with `model_format="str"`. (#87)
144+
145+
### Breaking changes
146+
147+
* The `.export()` method's `include` argument has been renamed to `content` (to match `.stream()`). (#75)
148+
149+
## [0.6.1] - 2025-04-03
150+
151+
### Bug fixes
152+
153+
* Fixed a missing dependency on the `requests` package.
154+
155+
## [0.6.0] - 2025-04-01
156+
157+
### New features
158+
159+
* New `content_pdf_file()` and `content_pdf_url()` allow you to upload PDFs to supported models. (#74)
160+
161+
### Improvements
162+
163+
* `Turn` and `Content` now inherit from `pydantic.BaseModel` to provide easier saving to and loading from JSON. (#72)
164+
165+
## [0.5.0] - 2025-03-18
166+
167+
### New features
168+
169+
* Added a `ChatSnowflake()` class to interact with [Snowflake Cortex LLM](https://docs.snowflake.com/en/user-guide/snowflake-cortex/llm-functions). (#54)
170+
* Added a `ChatAuto()` class, allowing for configuration of chat providers and models via environment variables. (#38, thanks @mconflitti-pbc)
171+
172+
### Improvements
173+
174+
* Updated `ChatAnthropic()`'s `model` default to `"claude-3-7-sonnet-latest"`. (#62)
175+
* The version is now accessible as `chatlas.__version__`. (#64)
176+
* All provider-specific `Chat` subclasses now have an associated extras in chatlas. For example, `ChatOpenAI` has `chatlas[openai]`, `ChatPerplexity` has `chatlas[perplexity]`, `ChatBedrockAnthropic` has `chatlas[bedrock-anthropic]`, and so forth for the other `Chat` classes. (#66)
177+
178+
### Bug fixes
179+
180+
* Fixed an issue with content getting duplicated when it overflows in a `Live()` console. (#71)
181+
* Fix an issue with tool calls not working with `ChatVertex()`. (#61)
182+
183+
184+
## [0.4.0] - 2025-02-19
185+
186+
### New features
187+
188+
* Added a `ChatVertex()` class to interact with Google Cloud's Vertex AI. (#50)
189+
* Added `.app(*, echo=)` support. This allows for chatlas to change the echo behavior when running the Shiny app. (#31)
190+
191+
### Improvements
192+
193+
* Migrated `ChatGoogle()`'s underlying python SDK from `google-generative` to `google-genai`. As a result, streaming tools are now working properly. (#50)
194+
195+
### Bug fixes
196+
197+
* Fixed a bug where synchronous chat tools would not work properly when used in a `_async()` context. (#56)
198+
* Fix broken `Chat`'s Shiny app when `.app(*, stream=True)` by using async chat tools. (#31)
199+
* Update formatting of exported markdown to use `repr()` instead of `str()` when exporting tool call results. (#30)
200+
201+
## [0.3.0] - 2024-12-20
202+
203+
### New features
204+
205+
* `Chat`'s `.tokens()` method gains a `values` argument. Set it to `"discrete"` to get a result that can be summed to determine the token cost of submitting the current turns. The default (`"cumulative"`), remains the same (the result can be summed to determine the overall token cost of the conversation).
206+
* `Chat` gains a `.token_count()` method to help estimate token cost of new input. (#23)
15207

16208
### Bug fixes
17209

18210
* `ChatOllama` no longer fails when a `OPENAI_API_KEY` environment variable is not set.
19211
* `ChatOpenAI` now correctly includes the relevant `detail` on `ContentImageRemote()` input.
212+
* `ChatGoogle` now correctly logs its `token_usage()`. (#23)
20213

21214

22215
## [0.2.0] - 2024-12-11
23216

24-
First stable release of `chatlas`, see the website to learn more <https://posit-dev.github.io/chatlas/>
217+
First stable release of `chatlas`, see the website to learn more <https://posit-dev.github.io/chatlas/>

0 commit comments

Comments
 (0)