o
~ri?4 @ s d dl mZ d dlZd dlZd dlmZ d dlmZmZm Z er&ddl
m Z G dd dZ G d d
d
e Z
G d d d e
ZG d
d de
ZdS ) ) annotationsN)Queue)
TYPE_CHECKINGAnycast )PreTrainedTokenizerBasec @ s e Zd ZdZdd Zdd ZdS ) BaseStreamerzG
Base class from which `.generate()` streamers should inherit.
c C t )z;Function that is called by `.generate()` to push new tokensNotImplementedErrorselfvalue r o/lsinfo/ai/hellotax_ai/llm_service/venv_embed/lib/python3.10/site-packages/transformers/generation/streamers.pyput zBaseStreamer.putc C r
)zHFunction that is called by `.generate()` to signal the end of generationr r r r r end$ r zBaseStreamer.endN)__name__
__module__ __qualname____doc__r r r r r r r s r c @ s@ e Zd ZdZddd d
Zd d Zd
d ZddddZdd ZdS ) TextStreamera)
Simple text streamer that prints the token(s) to stdout as soon as entire words are formed.
The API for the streamer classes is still under development and may change in the future.
Parameters:
tokenizer (`AutoTokenizer`):
The tokenized used to decode the tokens.
skip_prompt (`bool`, *optional*, defaults to `False`):
Whether to skip the prompt to `.generate()` or not. Useful e.g. for chatbots.
decode_kwargs (`dict`, *optional*):
Additional keyword arguments to pass to the tokenizer's `decode` method.
Examples:
```python
>>> from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer
>>> tok = AutoTokenizer.from_pretrained("openai-community/gpt2")
>>> model = AutoModelForCausalLM.from_pretrained("openai-community/gpt2")
>>> inputs = tok(["An increasing sequence: one,"], return_tensors="pt")
>>> streamer = TextStreamer(tok)
>>> # Despite returning the usual output, the streamer will also print the generated text to stdout.
>>> _ = model.generate(**inputs, streamer=streamer, max_new_tokens=20)
An increasing sequence: one, two, three, four, five, six, seven, eight, nine, ten, eleven,
```
F tokenizerr skip_promptbool
decode_kwargsr c K s( || _ || _|| _g | _d| _d| _d S )Nr T)r r r token_cache print_lennext_tokens_are_prompt)r r r r r r r __init__K s
zTextStreamer.__init__c C s t |jdkr|jd dkrtdt |jdkr |d }| jr(| jr(d| _dS | j| tt | j
j | jfi | j }|
drR|| jd }g | _d| _n5t |dkrr| t|d rr|| jd }| jt |7 _n|| j|dd }| jt |7 _| | dS ) zm
Receives tokens, decodes them, and prints them to stdout as soon as they form entire words.
r z'TextStreamer only supports batch size 1FN
)lenshape
ValueErrorr r! r extendtolistr strr decoder endswithr _is_chinese_charordrfindon_finalized_text)r r textprintable_textr r r r U s&
zTextStreamer.putc C sd t | jdkr$tt| jj| jfi | j}|| jd }g | _d| _nd}d| _| j |dd dS )z;Flushes any remaining cache and prints a newline to stdout.r N T)
stream_end)
r' r r r, r r- r r r! r2 )r r3 r4 r r r r w s zTextStreamer.endr3 r, r6 c C s t |d|sdndd dS )zNPrints the new text to stdout. If the stream is ending, also prints a newline.Tr5 N)flushr )printr r3 r6 r r r r2 s z TextStreamer.on_finalized_textc C s |dkr|dks@|dkr|dks@|dkr|dks@|dkr |dks@|d kr(|d
ks@|d kr0|d ks@|d
kr8|dks@|dkrB|dkrBdS dS )z6Checks whether CP is the codepoint of a CJK character.i N i i 4 iM i iߦ i i? i@ i i i i i i i TFr )r cpr r r r/ s z TextStreamer._is_chinese_charNF)r r r r r r r3 r, r6 r ) r r r r r" r r r2 r/ r r r r r ) s !
" r c D e Zd ZdZ dd fd d
ZddddZdd Zdd Z ZS )TextIteratorStreamera
Streamer that stores print-ready text in a queue, to be used by a downstream application as an iterator. This is
useful for applications that benefit from accessing the generated text in a non-blocking way (e.g. in an interactive
Gradio demo).
The API for the streamer classes is still under development and may change in the future.
Parameters:
tokenizer (`AutoTokenizer`):
The tokenized used to decode the tokens.
skip_prompt (`bool`, *optional*, defaults to `False`):
Whether to skip the prompt to `.generate()` or not. Useful e.g. for chatbots.
timeout (`float`, *optional*):
The timeout for the text queue. If `None`, the queue will block indefinitely. Useful to handle exceptions
in `.generate()`, when it is called in a separate thread.
decode_kwargs (`dict`, *optional*):
Additional keyword arguments to pass to the tokenizer's `decode` method.
Examples:
```python
>>> from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
>>> from threading import Thread
>>> tok = AutoTokenizer.from_pretrained("openai-community/gpt2")
>>> model = AutoModelForCausalLM.from_pretrained("openai-community/gpt2")
>>> inputs = tok(["An increasing sequence: one,"], return_tensors="pt")
>>> streamer = TextIteratorStreamer(tok)
>>> # Run the generation in a separate thread, so that we can fetch the generated text in a non-blocking way.
>>> generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=20)
>>> thread = Thread(target=model.generate, kwargs=generation_kwargs)
>>> thread.start()
>>> generated_text = ""
>>> for new_text in streamer:
... generated_text += new_text
>>> generated_text
'An increasing sequence: one, two, three, four, five, six, seven, eight, nine, ten, eleven,'
```
FNr r r r timeout float | Noner r c s. t j||fi | t | _d | _|| _d S N)superr" r
text_queue stop_signalr? )r r r r? r __class__r r r" s
z TextIteratorStreamer.__init__r3 r, r6 c C s2 | j j|| jd |r| j j| j| jd dS dS )\Put the new text in the queue. If the stream is ending, also put a stop signal in the queue.r? N)rC r r? rD r9 r r r r2 z&TextIteratorStreamer.on_finalized_textc C | S rA r r r r r __iter__ z TextIteratorStreamer.__iter__c C s$ | j j| jd}|| jkrt |S NrH )rC getr? rD
StopIterationr
r r r __next__ s
z TextIteratorStreamer.__next__FNr r r r r? r@ r r r; r< ) r r r r r" r2 rK rP
__classcell__r r rE r r> s 0 r> c r= )AsyncTextIteratorStreamera'
Streamer that stores print-ready text in a queue, to be used by a downstream application as an async iterator.
This is useful for applications that benefit from accessing the generated text asynchronously (e.g. in an
interactive Gradio demo).
The API for the streamer classes is still under development and may change in the future.
Parameters:
tokenizer (`AutoTokenizer`):
The tokenized used to decode the tokens.
skip_prompt (`bool`, *optional*, defaults to `False`):
Whether to skip the prompt to `.generate()` or not. Useful e.g. for chatbots.
timeout (`float`, *optional*):
The timeout for the text queue. If `None`, the queue will block indefinitely. Useful to handle exceptions
in `.generate()`, when it is called in a separate thread.
decode_kwargs (`dict`, *optional*):
Additional keyword arguments to pass to the tokenizer's `decode` method.
Raises:
TimeoutError: If token generation time exceeds timeout value.
Examples:
```python
>>> from transformers import AutoModelForCausalLM, AutoTokenizer, AsyncTextIteratorStreamer
>>> from threading import Thread
>>> import asyncio
>>> tok = AutoTokenizer.from_pretrained("openai-community/gpt2")
>>> model = AutoModelForCausalLM.from_pretrained("openai-community/gpt2")
>>> inputs = tok(["An increasing sequence: one,"], return_tensors="pt")
>>> # Run the generation in a separate thread, so that we can fetch the generated text in a non-blocking way.
>>> async def main():
... # Important: AsyncTextIteratorStreamer must be initialized inside a coroutine!
... streamer = AsyncTextIteratorStreamer(tok)
... generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=20)
... thread = Thread(target=model.generate, kwargs=generation_kwargs)
... thread.start()
... generated_text = ""
... async for new_text in streamer:
... generated_text += new_text
>>> print(generated_text)
>>> asyncio.run(main())
An increasing sequence: one, two, three, four, five, six, seven, eight, nine, ten, eleven,
```
FNr r r r r? r@ r r c sp t j||fi | t | _d | _|| _t | _t tdd }t
j dko)t || _
| j
r3|| _d S d | _d S )Nr? ) )rB r" asyncior rC rD r? get_running_looploopgetattrsys version_infocallablehas_asyncio_timeoutasyncio_timeout)r r r r? r timeout_contextrE r r r" " s
z"AsyncTextIteratorStreamer.__init__r3 r, r6 c C s2 | j | jj| |r| j | jj| j dS dS )rG N)rY call_soon_threadsaferC
put_nowaitrD r9 r r r r2 2 rI z+AsyncTextIteratorStreamer.on_finalized_textc C rJ rA r r r r r __aiter__8 rL z#AsyncTextIteratorStreamer.__aiter__c s zA| j r3| jd ur3| | j4 I d H | j I d H }W d I d H n 1 I d H s-w Y ntj| j | jdI d H }W n tjyM t w || jkrVt |S rM )
r^ r_ r? rC rN rW wait_for TimeoutErrorrD StopAsyncIterationr
r r r __anext__; s (
z#AsyncTextIteratorStreamer.__anext__rQ rR r; r< ) r r r r r" r2 rc rg rS r r rE r rT s 7 rT )
__future__r rW r[ queuer typingr r r tokenization_utils_baser r r r> rT r r r r s yK