Adding these to each generation chunk helps remove redundancy and
unecessary request ID operations.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
Doing this helps reduce the model's burden of generating the tool
call ID and type (which is always "function"). Follow mistral's spec
for tool call IDs by using a 9 character alphanumeric string.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
Re-rendering the template is an expensive operation when it's possible
to just concatenate the prompt and current generation text together.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
If a message with role = tool is present, the tool_call_id should
also be given to the template.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
When revisiting tool calls, the formats have more or less become standard.
For greater compatibility with templates, primarily use the message.tools
parameter and remove the extra custom metadata that is no longer required.
However, unlike other backends, tabbyAPI still uses template metadata
to declare what the tool start string is. This allows for template-level
customization along with giving more power to the user while the server
exists to consume rather than work on a case-by-case basis.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
To render in the template, tool call start tokens needed to have less
checks and remove the line to convert message.tool_calls to a dict
since that breaks the rest of the chain by disconnecting the types.
model_dump on the message itself already accomplishes this.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
It's useful for the client to know what the T/s and total time for
generation are per-request.
Works with both completions and chat completions.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
Matching YALS, if the model has add_bos_token enabled, then remove
an extra BOS token at the start of the prompt. This usually happens
with misconfigured templates such as Llama 3.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
Tools must be None by default. Chat completion message content can
be None, a string, or a list, so default to None. Exclude all None
values from a CC message since the template can say the variable
"exists" despite being None, causing an error.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
Messages were mistakenly being sent as Pydantic objects, but templates
expect dictionaries. Properly convert these before render.
In addition, initialize all Optional lists as an empty list since
this will cause the least problems when interacting with other parts
of API code, such as templates.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
When fetching special tokens from the model, don't factor in the
add_bos_token and ban_eos_token parameters as switches.
In addition, change the internal handling of add_bos_token to an optional
boolean. This allows us to fallback to the model when selecting whether
or not to add the BOS token, especially for chat completions.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
Since jobs are tracked via request IDs now, each generation task should
be uniquely identified in the event of cancellation.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
kwargs is pretty ugly when figuring out which arguments to use. The
base requests falls back to defaults anyways, so pass in the params
object as is.
However, since Python's typing isn't like TypeScript where types
can be transformed, the type hinting has a possiblity of None showing
up despite there always being a value for some params.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
* Add non-JSON version of `tools` and `functions` to `template_vars`.
Increase the compatibility with VLLM templates which use a non-JSON tools object.
* Add list of tool template variables to the documentation
* Use Jinja templates to provide `tools_json` and `functions_json`
This should be functionally equivelant, but the JSON won't be produced
unless it's needed.
* Make message.tool_calls match the JSON from ToolCallProcessor
* Log something when generating tool calls
* Add template for Qwen QwQ 32b
* Only log if tool calls have been detected
* API: Fix tool call variable assignments
Jinja functions do not run when variables are called. Use json.dumps
instead. In addition, log the request ID when stating that a tool
call was fired.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
* Add `ToolCallProcessor.dump()` to get the list of processed dicts
* Remove qwen_qwq_32b.jinja
This will be added to the following repository at a later date:
https://github.com/theroyallab/llm-prompt-templates
---------
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
Co-authored-by: kingbri <8082010+kingbri1@users.noreply.github.com>
* Fix Tool Call JSON Serialization Error
* Incorporate changes from PR 292
kingbri note: Adjusts the tool JSON formation and incorporates finish
reasons. Added both authors as co-authors due to edits on this commit
from the original PR.
Co-Authored-by: David Allada <dallada1@vt.edu>
Co-Authored-by: Benjamin Oldenburg <benjamin.oldenburg@ordis.co.th>
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
* API: Cleanup tool call JSON parsing
Split pre and post-processing of tool calls to its own class. This
cleans up the chat_completion utility module and also fixes the
JSON serialization bug.
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
---------
Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>
Co-authored-by: David Allada <dallada1@vt.edu>
Co-authored-by: kingbri <8082010+kingbri1@users.noreply.github.com>
The strings weren't being concatenated properly. Only add the combined
text if the chat completion type is a List.
Signed-off-by: kingbri <bdashore3@proton.me>
Previously, the flow for parsing chat completion messages and rendering
from the prompt template was disconnected between endpoints. Now, create
a common function to render and handle everything appropriately afterwards.
Signed-off-by: kingbri <bdashore3@proton.me>
Migrate the add method into the class itself. Also, a BaseModel isn't
needed here since this isn't a serialized class.
Signed-off-by: kingbri <bdashore3@proton.me>
Previously, the messages were a list of dicts. These are untyped
and don't provide strict hinting. Add types for chat completion
messages and reformat existing code.
Signed-off-by: kingbri <bdashore3@proton.me>
* Support image_url inputs containing URLs or base64 strings following OAI vision spec
* Use async lru cache for image embeddings
* Add generic wrapper class for multimodal embeddings
* improve validation
* remove to_gen_params functions
* update changes for all endpoint types
* OAI: Fix calls to generation
Chat completion and completion need to have prompt split out before
pushing to the backend.
Signed-off-by: kingbri <bdashore3@proton.me>
* Sampling: Convert Top-K values of -1 to 0
Some OAI implementations use -1 as disabled instead of 0. Therefore,
add a coalesce case.
Signed-off-by: kingbri <bdashore3@proton.me>
* Sampling: Format and space out
Make the code more readable.
Signed-off-by: kingbri <bdashore3@proton.me>
* Sampling: Fix mirostat
Field items are nested in data within a Pydantic FieldInfo
Signed-off-by: kingbri <bdashore3@proton.me>
* Sampling: Format
Signed-off-by: kingbri <bdashore3@proton.me>
* Sampling: Fix banned_tokens and allowed_tokens conversion
If the provided string has whitespace, trim it before splitting.
Signed-off-by: kingbri <bdashore3@proton.me>
* Sampling: Add helpful log to dry_sequence_breakers
Let the user know if the sequence errors out.
Signed-off-by: kingbri <bdashore3@proton.me>
* Sampling: Apply validators in right order
Validators need to be applied in order from top to bottom, this is why
the after validator was not being applied properly.
Set the model to validate default params for sampler override purposes.
This can be turned off if there are unclear errors.
Signed-off-by: kingbri <bdashore3@proton.me>
* Endpoints: Format
Cleanup and semantically fix field validators
Signed-off-by: kingbri <bdashore3@proton.me>
* Kobold: Update validators and fix parameter application
Validators on parent fields cannot see child fields. Therefore,
validate using the child fields instead and alter the parent field
data from there.
Also fix badwordsids casting.
Signed-off-by: kingbri <bdashore3@proton.me>
* Sampling: Remove validate defaults and fix mirostat
If a user sets an override to a non-default value, that's their
own fault.
Run validator on the actual mirostat_mode parameter rather than
the alternate mirostat parameter.
Signed-off-by: kingbri <bdashore3@proton.me>
* Kobold: Rework badwordsids
Currently, this serves to ban the EOS token. All other functionality
was legacy, so remove it.
Signed-off-by: kingbri <bdashore3@proton.me>
* Model: Remove HuggingfaceConfig
This was only necessary for badwordsids. All other fields are handled
by exl2. Keep the class as a stub if it's needed again.
Signed-off-by: kingbri <bdashore3@proton.me>
* Kobold: Bump kcpp impersonation
TabbyAPI supports XTC now.
Signed-off-by: kingbri <bdashore3@proton.me>
* Sampling: Change alias to validation_alias
Reduces the probability for errors and makes the class consistent.
Signed-off-by: kingbri <bdashore3@proton.me>
* OAI: Use constraints for validation
Instead of adding a model_validator, use greater than or equal to
constraints provided by Pydantic.
Signed-off-by: kingbri <bdashore3@proton.me>
* Tree: Lint
Signed-off-by: kingbri <bdashore3@proton.me>
---------
Co-authored-by: SecretiveShell <84923604+SecretiveShell@users.noreply.github.com>
Co-authored-by: kingbri <bdashore3@proton.me>
Make it so any message role can be parsed from a list. Not really
sure why this is the case because system and assistant shouldn't be
sending data other than text, but it also doesn't make much sense
to be extremely strict with roles either.
Signed-off-by: kingbri <bdashore3@proton.me>
Metadata is generated via a template's module. This requires a single
iteration through the template. If a template tries to access a passed
variable that doesn't exist, it will error.
Therefore, generate the metadata at runtime to prevent these errors
from happening. To optimize further, cache the metadata after the
first generation to prevent the expensive call of making a template
module.
Signed-off-by: kingbri <bdashore3@proton.me>
* returning stop str if exists from gen
* added chat template for firefunctionv2
* pulling tool vars from template
* adding parsing for tool inputs/outputs
* passing tool data from endpoint to chat template, adding tool_start to the stop list
* loosened typing on the response tool call, leaning more on the user supplying a quality schema if they want a particular format
* non streaming generation prototype
* cleaning template
* Continued work with type, ingestion into template, and chat template for fire func
* Correction - streaming toolcall comes back as delta obj not inside chatcomprespchoice per chat_completion_chunk.py inside OAI lib.
* Ruff Formating
* Moved stop string and tool updates out of prompt creation func
Updated tool pydantic to match OAI
Support for streaming
Updated generate tool calls to use flag within chat_template and insert tool reminder
* Llama 3.1 chat templates
Updated fire func template
* renamed llama3.1 to chatml_with_headers..
* update name of template
* Support for calling a tool start token rather than the string.
Simplified tool_params
Warning when gen_settings are being overidden becuase user set temp to 0
Corrected schema and tools to correct types for function args. Str for some reason
* draft groq tool use model template
* changed headers to vars for readablity (but mostly because some models are weird about newlines after headers, so this is an easier way to change globally)
* Clean up comments and code in chat comp
* Post processed tool call to meet OAI spec rather than forcing model to write json in a string in the middle of the call.
* changes example back to args as json rather than string of json
* Standardize chat templates to each other
* cleaning/rewording
* stop elements can also be ints (tokens)
* Cleaning/formatting
* added special tokens for tools and tool_response as specified in description
* Cleaning
* removing aux templates - going to live in llm-promp-templates repo instead
* Tree: Format
Signed-off-by: kingbri <bdashore3@proton.me>
* Chat Completions: Don't include internal tool variables in OpenAPI
Use SkipJsonSchema to supress inclusion with the OpenAPI JSON. The
location of these variables may need to be changed in the future.
Signed-off-by: kingbri <bdashore3@proton.me>
* Templates: Deserialize metadata on template load
Since we're only looking for specific template variables that are
static in the template, it makes more sense to render when the template
is initialized.
Signed-off-by: kingbri <bdashore3@proton.me>
* Tools: Fix comments
Adhere to the format style of comments in the rest of the project.
Signed-off-by: kingbri <bdashore3@proton.me>
---------
Co-authored-by: Ben Gitter <gitterbd@gmail.com>
Signed-off-by: kingbri <bdashore3@proton.me>
Uvicorn can log in both the request disconnect handler and the
CancelledError. However, these sometimes don't work and both
need to be checked. But, don't log twice if one works.
Signed-off-by: kingbri <bdashore3@proton.me>
Identify which request is being processed to help users disambiguate
which logs correspond to which request.
Signed-off-by: kingbri <bdashore3@proton.me>
Place the logic into their proper utility functions and cleanup
the code with formatting.
Also, OAI's docs specify that a [DONE] return is needed when everything
is finished.
Signed-off-by: kingbri <bdashore3@proton.me>
Use a queue-based system to get choices independently and send them
in the overall streaming payload. This method allows for unordered
streaming of generations.
The system is a bit redundant, so maybe make the code more optimized
in the future.
Signed-off-by: kingbri <bdashore3@proton.me>
For multiple generations in the same request, nested arrays kept their
original reference, resulting in duplications. This will occur with
any collection type.
For optimization purposes, a deepcopy isn't run for the first iteration
since original references are created.
This is not the most elegant solution, but it works for the described
cases.
Signed-off-by: kingbri <bdashore3@proton.me>
This adds the ability to add multiple choices to a generation. This
is only available for non-streaming gens for now, it requires some
more work to port over to streaming.
Signed-off-by: kingbri <bdashore3@proton.me>
Waiting for request disconnect takes some extra time and allows
generation chunks to pile up, resulting in large payloads being sent
at once not making up a smooth stream.
Use the polling method in non-streaming requests by creating a background
task and then check if the task is done, signifying that the request
has been disconnected.
Signed-off-by: kingbri <bdashore3@proton.me>
Depending on the day of the week, Starlette can work with a CancelledError
or using await request.is_disconnected(). Run the same behavior for both
cases and allow cancellation.
Streaming requests now set an event to cancel the batched job and break
out of the generation loop.
Signed-off-by: kingbri <bdashore3@proton.me>
The new async dynamic job allows for native async support without the
need of threading. Also add logprobs and metrics back to responses.
Signed-off-by: kingbri <bdashore3@proton.me>
response_prefix is used to add a prefix before generating the next
message. This is used in many cases such as continuining a prompt
(see #96).
Also if a template has BOS token specified, add_bos_token will
append two BOS tokens. Add a check which strips a starting BOS token
from the prompt if it exists.
Signed-off-by: kingbri <bdashore3@proton.me>
Having many utility functions for initialization doesn't make much sense.
Instead, handle anything regarding template creation inside the
class which reduces the amount of function imports.
Signed-off-by: kingbri <bdashore3@proton.me>
