From b3cde9900c55a73f8a28044d2a4da91a70bbf3a0 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Fri, 18 Oct 2024 08:21:54 +0800
Subject: [PATCH] feat: add parameter top-k for the llm model provided by
 openrouter and siliconflow (#9455)

---
 .../model_providers/openrouter/llm/deepseek-chat.yaml    | 9 +++++++++
 .../model_providers/openrouter/llm/deepseek-coder.yaml   | 9 +++++++++
 .../model_providers/openrouter/llm/gpt-3.5-turbo.yaml    | 9 +++++++++
 .../model_providers/openrouter/llm/gpt-4-32k.yaml        | 9 +++++++++
 .../model_providers/openrouter/llm/gpt-4.yaml            | 9 +++++++++
 .../openrouter/llm/gpt-4o-2024-08-06.yaml                | 9 +++++++++
 .../model_providers/openrouter/llm/gpt-4o-mini.yaml      | 9 +++++++++
 .../model_providers/openrouter/llm/gpt-4o.yaml           | 9 +++++++++
 .../openrouter/llm/llama-3-70b-instruct.yaml             | 9 +++++++++
 .../openrouter/llm/llama-3-8b-instruct.yaml              | 9 +++++++++
 .../openrouter/llm/llama-3.1-405b-instruct.yaml          | 9 +++++++++
 .../openrouter/llm/llama-3.1-70b-instruct.yaml           | 9 +++++++++
 .../openrouter/llm/llama-3.1-8b-instruct.yaml            | 9 +++++++++
 .../openrouter/llm/mistral-7b-instruct.yaml              | 9 +++++++++
 .../openrouter/llm/mixtral-8x22b-instruct.yaml           | 9 +++++++++
 .../openrouter/llm/mixtral-8x7b-instruct.yaml            | 9 +++++++++
 .../model_providers/openrouter/llm/o1-mini.yaml          | 9 +++++++++
 .../model_providers/openrouter/llm/o1-preview.yaml       | 9 +++++++++
 .../openrouter/llm/qwen2-72b-instruct.yaml               | 9 +++++++++
 .../openrouter/llm/qwen2.5-72b-instruct.yaml             | 9 +++++++++
 .../siliconflow/llm/deepdeek-coder-v2-instruct.yaml      | 9 +++++++++
 .../siliconflow/llm/deepseek-v2-chat.yaml                | 9 +++++++++
 .../model_providers/siliconflow/llm/deepseek-v2.5.yaml   | 9 +++++++++
 .../model_providers/siliconflow/llm/gemma-2-27b-it.yaml  | 9 +++++++++
 .../model_providers/siliconflow/llm/gemma-2-9b-it.yaml   | 9 +++++++++
 .../model_providers/siliconflow/llm/glm4-9b-chat.yaml    | 9 +++++++++
 .../siliconflow/llm/internlm2_5-20b-chat.yaml            | 9 +++++++++
 .../siliconflow/llm/internlm2_5-7b-chat.yaml             | 9 +++++++++
 .../siliconflow/llm/meta-mlama-3-70b-instruct.yaml       | 9 +++++++++
 .../siliconflow/llm/meta-mlama-3-8b-instruct.yaml        | 9 +++++++++
 .../siliconflow/llm/meta-mlama-3.1-405b-instruct.yaml    | 9 +++++++++
 .../siliconflow/llm/meta-mlama-3.1-70b-instruct.yaml     | 9 +++++++++
 .../siliconflow/llm/meta-mlama-3.1-8b-instruct.yaml      | 9 +++++++++
 .../siliconflow/llm/mistral-7b-instruct-v0.2.yaml        | 9 +++++++++
 .../siliconflow/llm/mistral-8x7b-instruct-v0.1.yaml      | 9 +++++++++
 .../siliconflow/llm/qwen2-1.5b-instruct.yaml             | 9 +++++++++
 .../siliconflow/llm/qwen2-57b-a14b-instruct.yaml         | 9 +++++++++
 .../siliconflow/llm/qwen2-72b-instruct.yaml              | 9 +++++++++
 .../siliconflow/llm/qwen2-7b-instruct.yaml               | 9 +++++++++
 .../siliconflow/llm/qwen2.5-14b-instruct.yaml            | 9 +++++++++
 .../siliconflow/llm/qwen2.5-32b-instruct.yaml            | 9 +++++++++
 .../siliconflow/llm/qwen2.5-72b-instruct.yaml            | 9 +++++++++
 .../siliconflow/llm/qwen2.5-7b-instruct.yaml             | 9 +++++++++
 .../model_providers/siliconflow/llm/yi-1.5-34b-chat.yaml | 9 +++++++++
 .../model_providers/siliconflow/llm/yi-1.5-6b-chat.yaml  | 9 +++++++++
 .../model_providers/siliconflow/llm/yi-1.5-9b-chat.yaml  | 9 +++++++++
 46 files changed, 414 insertions(+)

diff --git a/api/core/model_runtime/model_providers/openrouter/llm/deepseek-chat.yaml b/api/core/model_runtime/model_providers/openrouter/llm/deepseek-chat.yaml
index 7a1dea6950..6743bfcad6 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/deepseek-chat.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/deepseek-chat.yaml
@@ -35,6 +35,15 @@ parameter_rules:
     help:
       zh_Hans: 控制生成结果的随机性。数值越小，随机性越弱；数值越大，随机性越强。一般而言，top_p 和 temperature 两个参数选择一个进行调整即可。
       en_US: Control the randomness of generated results. The smaller the value, the weaker the randomness; the larger the value, the stronger the randomness. Generally speaking, you can adjust one of the two parameters top_p and temperature.
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
     default: 0
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/deepseek-coder.yaml b/api/core/model_runtime/model_providers/openrouter/llm/deepseek-coder.yaml
index c05f4769b8..375a4d2d52 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/deepseek-coder.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/deepseek-coder.yaml
@@ -18,6 +18,15 @@ parameter_rules:
     min: 0
     max: 1
     default: 1
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: max_tokens
     use_template: max_tokens
     min: 1
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/gpt-3.5-turbo.yaml b/api/core/model_runtime/model_providers/openrouter/llm/gpt-3.5-turbo.yaml
index 186c1cc663..621ecf065e 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/gpt-3.5-turbo.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/gpt-3.5-turbo.yaml
@@ -14,6 +14,15 @@ parameter_rules:
     use_template: temperature
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: presence_penalty
     use_template: presence_penalty
   - name: frequency_penalty
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/gpt-4-32k.yaml b/api/core/model_runtime/model_providers/openrouter/llm/gpt-4-32k.yaml
index 8c2989b300..887e6d60f9 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/gpt-4-32k.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/gpt-4-32k.yaml
@@ -14,6 +14,15 @@ parameter_rules:
     use_template: temperature
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: presence_penalty
     use_template: presence_penalty
   - name: frequency_penalty
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/gpt-4.yaml b/api/core/model_runtime/model_providers/openrouter/llm/gpt-4.yaml
index ef19d4f6f0..66d1f9ae67 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/gpt-4.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/gpt-4.yaml
@@ -14,6 +14,15 @@ parameter_rules:
     use_template: temperature
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: presence_penalty
     use_template: presence_penalty
   - name: frequency_penalty
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/gpt-4o-2024-08-06.yaml b/api/core/model_runtime/model_providers/openrouter/llm/gpt-4o-2024-08-06.yaml
index 0be325f55b..695cc3eedf 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/gpt-4o-2024-08-06.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/gpt-4o-2024-08-06.yaml
@@ -16,6 +16,15 @@ parameter_rules:
     use_template: temperature
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: presence_penalty
     use_template: presence_penalty
   - name: frequency_penalty
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/gpt-4o-mini.yaml b/api/core/model_runtime/model_providers/openrouter/llm/gpt-4o-mini.yaml
index 3b1d95643d..e1e5889085 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/gpt-4o-mini.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/gpt-4o-mini.yaml
@@ -15,6 +15,15 @@ parameter_rules:
     use_template: temperature
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: presence_penalty
     use_template: presence_penalty
   - name: frequency_penalty
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/gpt-4o.yaml b/api/core/model_runtime/model_providers/openrouter/llm/gpt-4o.yaml
index a8c97efdd6..560bf9d7d0 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/gpt-4o.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/gpt-4o.yaml
@@ -15,6 +15,15 @@ parameter_rules:
     use_template: temperature
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: presence_penalty
     use_template: presence_penalty
   - name: frequency_penalty
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/llama-3-70b-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/llama-3-70b-instruct.yaml
index b91c39e729..04a4a90c6d 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/llama-3-70b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/llama-3-70b-instruct.yaml
@@ -10,6 +10,15 @@ parameter_rules:
     use_template: temperature
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: max_tokens
     use_template: max_tokens
     required: true
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/llama-3-8b-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/llama-3-8b-instruct.yaml
index 84b2c7fac2..066949d431 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/llama-3-8b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/llama-3-8b-instruct.yaml
@@ -10,6 +10,15 @@ parameter_rules:
     use_template: temperature
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: max_tokens
     use_template: max_tokens
     required: true
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/llama-3.1-405b-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/llama-3.1-405b-instruct.yaml
index a489ce1b5a..0cd89dea71 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/llama-3.1-405b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/llama-3.1-405b-instruct.yaml
@@ -10,6 +10,15 @@ parameter_rules:
     use_template: temperature
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: max_tokens
     use_template: max_tokens
     required: true
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/llama-3.1-70b-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/llama-3.1-70b-instruct.yaml
index 12037411b1..768ab5ecbb 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/llama-3.1-70b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/llama-3.1-70b-instruct.yaml
@@ -10,6 +10,15 @@ parameter_rules:
     use_template: temperature
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: max_tokens
     use_template: max_tokens
     required: true
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/llama-3.1-8b-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/llama-3.1-8b-instruct.yaml
index 6f06493f29..67b6b82b5d 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/llama-3.1-8b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/llama-3.1-8b-instruct.yaml
@@ -10,6 +10,15 @@ parameter_rules:
     use_template: temperature
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: max_tokens
     use_template: max_tokens
     required: true
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/mistral-7b-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/mistral-7b-instruct.yaml
index 012dfc55ce..d08c016e95 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/mistral-7b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/mistral-7b-instruct.yaml
@@ -18,6 +18,15 @@ parameter_rules:
     default: 1
     min: 0
     max: 1
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: max_tokens
     use_template: max_tokens
     default: 1024
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/mixtral-8x22b-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/mixtral-8x22b-instruct.yaml
index f4eb4e45d9..e3af0e64d8 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/mixtral-8x22b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/mixtral-8x22b-instruct.yaml
@@ -18,6 +18,15 @@ parameter_rules:
     default: 1
     min: 0
     max: 1
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: max_tokens
     use_template: max_tokens
     default: 1024
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/mixtral-8x7b-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/mixtral-8x7b-instruct.yaml
index 7871e1f7a0..095ea5a858 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/mixtral-8x7b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/mixtral-8x7b-instruct.yaml
@@ -19,6 +19,15 @@ parameter_rules:
     default: 1
     min: 0
     max: 1
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: max_tokens
     use_template: max_tokens
     default: 1024
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/o1-mini.yaml b/api/core/model_runtime/model_providers/openrouter/llm/o1-mini.yaml
index 85a918ff5e..f4202ee814 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/o1-mini.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/o1-mini.yaml
@@ -12,6 +12,15 @@ parameter_rules:
     use_template: temperature
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: presence_penalty
     use_template: presence_penalty
   - name: frequency_penalty
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/o1-preview.yaml b/api/core/model_runtime/model_providers/openrouter/llm/o1-preview.yaml
index 74b0a511be..1281b84286 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/o1-preview.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/o1-preview.yaml
@@ -12,6 +12,15 @@ parameter_rules:
     use_template: temperature
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: presence_penalty
     use_template: presence_penalty
   - name: frequency_penalty
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/qwen2-72b-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/qwen2-72b-instruct.yaml
index 7b75fcb0c9..b6058138d3 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/qwen2-72b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/qwen2-72b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/qwen2.5-72b-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/qwen2.5-72b-instruct.yaml
index f141a40a00..5392b11168 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/qwen2.5-72b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/qwen2.5-72b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/deepdeek-coder-v2-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/deepdeek-coder-v2-instruct.yaml
index d4431179e5..d5f23776ea 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/deepdeek-coder-v2-instruct.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/deepdeek-coder-v2-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/deepseek-v2-chat.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/deepseek-v2-chat.yaml
index caa6508b5e..7aa684ef38 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/deepseek-v2-chat.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/deepseek-v2-chat.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/deepseek-v2.5.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/deepseek-v2.5.yaml
index 1c8e15ae52..b30fa3e2d1 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/deepseek-v2.5.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/deepseek-v2.5.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/gemma-2-27b-it.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/gemma-2-27b-it.yaml
index 2840e3dcf4..f2a1f64bfb 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/gemma-2-27b-it.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/gemma-2-27b-it.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/gemma-2-9b-it.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/gemma-2-9b-it.yaml
index d7e19b46f6..b096b9b647 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/gemma-2-9b-it.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/gemma-2-9b-it.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/glm4-9b-chat.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/glm4-9b-chat.yaml
index 9b32a02477..87acc557b7 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/glm4-9b-chat.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/glm4-9b-chat.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/internlm2_5-20b-chat.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/internlm2_5-20b-chat.yaml
index d9663582e5..60157c2b46 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/internlm2_5-20b-chat.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/internlm2_5-20b-chat.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/internlm2_5-7b-chat.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/internlm2_5-7b-chat.yaml
index 73ad4480aa..faf4af7ea3 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/internlm2_5-7b-chat.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/internlm2_5-7b-chat.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3-70b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3-70b-instruct.yaml
index 9993d781ac..d01770cb01 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3-70b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3-70b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3-8b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3-8b-instruct.yaml
index 60e3764789..3cd75d89e8 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3-8b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3-8b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3.1-405b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3.1-405b-instruct.yaml
index f992660aa2..3506a70bcc 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3.1-405b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3.1-405b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3.1-70b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3.1-70b-instruct.yaml
index 1c69d63a40..994a754a82 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3.1-70b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3.1-70b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3.1-8b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3.1-8b-instruct.yaml
index a97002a5ca..ebfa9aac9d 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3.1-8b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/meta-mlama-3.1-8b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/mistral-7b-instruct-v0.2.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/mistral-7b-instruct-v0.2.yaml
index 89fb153ba0..a71d8688a8 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/mistral-7b-instruct-v0.2.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/mistral-7b-instruct-v0.2.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/mistral-8x7b-instruct-v0.1.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/mistral-8x7b-instruct-v0.1.yaml
index 2785e7496f..db45a75c6d 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/mistral-8x7b-instruct-v0.1.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/mistral-8x7b-instruct-v0.1.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-1.5b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-1.5b-instruct.yaml
index f6c976af8e..bec5d37c57 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-1.5b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-1.5b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-57b-a14b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-57b-a14b-instruct.yaml
index a996e919ea..b2461335f8 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-57b-a14b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-57b-a14b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-72b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-72b-instruct.yaml
index a6e2c22dac..e0f23bd89e 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-72b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-72b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-7b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-7b-instruct.yaml
index d8bea5e129..47a9da8119 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-7b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2-7b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-14b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-14b-instruct.yaml
index 02a401464b..9cc5ac4c91 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-14b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-14b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-32b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-32b-instruct.yaml
index d084617e7d..c7fb21e9e1 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-32b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-32b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-72b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-72b-instruct.yaml
index dfbad2494c..03136c88a1 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-72b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-72b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-7b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-7b-instruct.yaml
index cdc8ffc4d2..99412adde7 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-7b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-7b-instruct.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/yi-1.5-34b-chat.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/yi-1.5-34b-chat.yaml
index 864ba46f1a..3e25f82369 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/yi-1.5-34b-chat.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/yi-1.5-34b-chat.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/yi-1.5-6b-chat.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/yi-1.5-6b-chat.yaml
index fe4c8b4b3e..827b2ce1e5 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/yi-1.5-6b-chat.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/yi-1.5-6b-chat.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing:
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/yi-1.5-9b-chat.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/yi-1.5-9b-chat.yaml
index c61f0dc53f..112fcbfe97 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/yi-1.5-9b-chat.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/yi-1.5-9b-chat.yaml
@@ -21,6 +21,15 @@ parameter_rules:
       en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
   - name: top_p
     use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
   - name: frequency_penalty
     use_template: frequency_penalty
 pricing: