apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-05 00:33:03 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

fix topp default value (#2814)
Some checks failed
Deploy GitHub Pages / deploy (push) Has been cancelled
Details

Browse Source
This commit is contained in:
Sunny-bot1
2025-07-11 17:10:21 +08:00
committed by GitHub
parent c08561c13a
commit f6ad26fc08
5 changed files with 9 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
fastdeploy/engine/sampling_params.py
View File

@@ -82,7 +82,7 @@ class SamplingParams:
frequency_penalty: float = None
repetition_penalty: float = None
temperature: float = None
top_p: float = 1.0
top_p: float = None
top_k: int = 0
seed: Optional[int] = None
stop: Optional[Union[str, List[str]]] = None
@@ -132,7 +132,7 @@ class SamplingParams:
repetition_penalty=repetition_penalty
if repetition_penalty is not None else 1.0,
temperature=temperature if temperature is not None else 1.0,
top_p=top_p if top_p is not None else 1.0,
top_p=top_p,
top_k=top_k if top_k is not None else 0,
seed=seed,
stop=stop,

4
fastdeploy/worker/gcu_model_runner.py
View File

@@ -24,6 +24,7 @@ from paddleformers.utils.log import logger
from fastdeploy.config import FDConfig
from fastdeploy.engine.request import Request
from fastdeploy.model_executor.forward_meta import ForwardMeta
from fastdeploy.model_executor.guided_decoding import get_guided_backend
from fastdeploy.model_executor.guided_decoding.base_guided_decoding import \
LogitsProcessorBase
@@ -39,7 +40,6 @@ from fastdeploy.model_executor.ops.gcu import set_value_by_flags_and_idx
from fastdeploy.model_executor.pre_and_post_process import (post_process,
pre_process,
rebuild_padding)
from fastdeploy.model_executor.forward_meta import ForwardMeta
from fastdeploy.worker.model_runner_base import ModelRunnerBase
from fastdeploy.worker.output import ModelOutputData, ModelRunnerOutput
@@ -235,7 +235,7 @@ class GCUModelRunner(ModelRunnerBase):
request.eos_token_ids.append(request.eos_token_ids[0])
self.share_inputs["eos_token_id"][:] = np.array(
request.eos_token_ids, dtype="int64").reshape(-1, 1)
self.share_inputs["top_p"][idx:idx + 1] = request.get("top_p", 1.0)
self.share_inputs["top_p"][idx:idx + 1] = request.get("top_p", 0.7)
self.share_inputs["top_k"][idx:idx + 1] = request.get("top_k", 0)
self.share_inputs["temperature"][idx:idx + 1] = request.get(
"temperature", 0.95)

2
fastdeploy/worker/gpu_model_runner.py
View File

@@ -245,7 +245,7 @@ class GPUModelRunner(ModelRunnerBase):
request.eos_token_ids.append(request.eos_token_ids[0])
self.share_inputs["eos_token_id"][:] = np.array(
request.eos_token_ids, dtype="int64").reshape(-1, 1)
self.share_inputs["top_p"][idx:idx + 1] = request.get("top_p", 1.0)
self.share_inputs["top_p"][idx:idx + 1] = request.get("top_p", 0.7)
self.share_inputs["top_k"][idx:idx + 1] = request.get("top_k", 0)
self.share_inputs["temperature"][idx:idx + 1] = request.get(
"temperature", 0.95)

4
fastdeploy/worker/iluvatar_model_runner.py
View File

@@ -24,6 +24,7 @@ from paddleformers.utils.log import logger
from fastdeploy.config import FDConfig
from fastdeploy.engine.request import Request
from fastdeploy.model_executor.forward_meta import ForwardMeta
from fastdeploy.model_executor.layers.attention import get_attention_backend
from fastdeploy.model_executor.layers.attention.base_attention_backend import \
AttentionBackend
@@ -37,7 +38,6 @@ from fastdeploy.model_executor.pre_and_post_process import (post_process,
pre_process,
rebuild_padding,
step_cuda)
from fastdeploy.model_executor.forward_meta import ForwardMeta
from fastdeploy.worker.model_runner_base import ModelRunnerBase
from fastdeploy.worker.output import ModelOutputData, ModelRunnerOutput
@@ -224,7 +224,7 @@ class IluvatarModelRunner(ModelRunnerBase):
request.eos_token_ids.append(request.eos_token_ids[0])
self.share_inputs["eos_token_id"][:] = np.array(
request.eos_token_ids, dtype="int64").reshape(-1, 1)
self.share_inputs["top_p"][idx:idx + 1] = request.get("top_p", 1.0)
self.share_inputs["top_p"][idx:idx + 1] = request.get("top_p", 0.7)
self.share_inputs["top_k"][idx:idx + 1] = request.get("top_k", 0)
self.share_inputs["temperature"][idx:idx + 1] = request.get(
"temperature", 0.95)

4
fastdeploy/worker/xpu_model_runner.py
View File

@@ -23,6 +23,7 @@ import paddle.nn as nn
from fastdeploy.config import FDConfig
from fastdeploy.engine.request import Request
from fastdeploy.model_executor.forward_meta import ForwardMeta, XPUForwardMeta
from fastdeploy.model_executor.layers.attention import get_attention_backend
from fastdeploy.model_executor.layers.attention.base_attention_backend import \
AttentionBackend
@@ -31,7 +32,6 @@ from fastdeploy.model_executor.layers.sample.meta_data import SamplingMetadata
from fastdeploy.model_executor.layers.sample.sampler import Sampler
from fastdeploy.model_executor.model_loader import get_model_from_loader
from fastdeploy.utils import get_logger
from fastdeploy.model_executor.forward_meta import ForwardMeta, XPUForwardMeta
from fastdeploy.worker.model_runner_base import ModelRunnerBase
from fastdeploy.worker.output import ModelOutputData, ModelRunnerOutput
@@ -295,7 +295,7 @@ class XPUModelRunner(ModelRunnerBase):
self.share_inputs["eos_token_id"][:] = np.array(
request.eos_token_ids, dtype="int64").reshape(-1, 1)
self.share_inputs["pre_ids"][idx:idx + 1] = -1
self.share_inputs["top_p"][idx:idx + 1] = request.get("top_p", 1.0)
self.share_inputs["top_p"][idx:idx + 1] = request.get("top_p", 0.7)
self.share_inputs["top_k"][idx:idx + 1] = request.get("top_k", 0)
self.share_inputs["temperature"][idx:idx + 1] = request.get(
"temperature", 0.95)