add temperature for system prompt

src/distilabel_dataset_generator/apps/sft.py

@@ -53,11 +53,11 @@ def convert_dataframe_messages(dataframe: pd.DataFrame) -> pd.DataFrame:
     return dataframe
 
 
-def generate_system_prompt(dataset_description, progress=gr.Progress()):
+def generate_system_prompt(dataset_description, temperature, progress=gr.Progress()):
     progress(0.0, desc="Generating system prompt")
 
     progress(0.3, desc="Initializing text generation")
-    generate_description = get_prompt_generator()
+    generate_description = get_prompt_generator(temperature)
     progress(0.7, desc="Generating system prompt")
     result = next(
         generate_description.process(
@@ -360,6 +360,15 @@ with gr.Blocks(css=_LOGGED_OUT_CSS) as app:
                     label="Dataset description",
                     placeholder="Give a precise description of your desired dataset.",
                 )
+                with gr.Accordion("Temperature", open=False):
+                    temperature = gr.Slider(
+                        minimum=0.1,
+                        maximum=1,
+                        value=0.8,
+                        step=0.1,
+                        interactive=True,
+                        show_label=False,
+                    )
                 load_btn = gr.Button(
                     "Create dataset",
                     variant="primary",
@@ -444,7 +453,7 @@ with gr.Blocks(css=_LOGGED_OUT_CSS) as app:
     gr.on(
         triggers=[load_btn.click, btn_apply_to_sample_dataset.click],
         fn=generate_system_prompt,
-        inputs=[dataset_description],
+        inputs=[dataset_description, temperature],
         outputs=[system_prompt, dataframe],
         show_progress=True,
     ).then(

src/distilabel_dataset_generator/apps/textcat.py

@@ -39,10 +39,10 @@ from src.distilabel_dataset_generator.utils import (
 )
 
 
-def generate_system_prompt(dataset_description, progress=gr.Progress()):
+def generate_system_prompt(dataset_description, temperature, progress=gr.Progress()):
     progress(0.0, desc="Generating text classification task")
     progress(0.3, desc="Initializing text generation")
-    generate_description = get_prompt_generator()
+    generate_description = get_prompt_generator(temperature)
     progress(0.7, desc="Generating text classification task")
     system_prompt = next(
         generate_description.process(
@@ -368,6 +368,15 @@ with gr.Blocks(css=_LOGGED_OUT_CSS) as app:
                     label="Dataset description",
                     placeholder="Give a precise description of your desired dataset.",
                 )
+                with gr.Accordion("Temperature", open=False):
+                    temperature = gr.Slider(
+                        minimum=0.1,
+                        maximum=1,
+                        value=0.8,
+                        step=0.1,
+                        interactive=True,
+                        show_label=False,
+                    )
                 load_btn = gr.Button(
                     "Create dataset",
                     variant="primary",
@@ -490,7 +499,7 @@ with gr.Blocks(css=_LOGGED_OUT_CSS) as app:
     gr.on(
         triggers=[load_btn.click, btn_apply_to_sample_dataset.click],
         fn=generate_system_prompt,
-        inputs=[dataset_description],
+        inputs=[dataset_description, temperature],
         outputs=[system_prompt, dataframe],
         show_progress=True,
     ).then(

src/distilabel_dataset_generator/pipelines/sft.py

@@ -183,6 +183,24 @@ if __name__ == "__main__":
     return code
 
 
+def get_prompt_generator(temperature):
+    prompt_generator = TextGeneration(
+        llm=InferenceEndpointsLLM(
+            api_key=_get_next_api_key(),
+            model_id=MODEL,
+            tokenizer_id=MODEL,
+            generation_kwargs={
+                "temperature": temperature,
+                "max_new_tokens": 2048,
+                "do_sample": True,
+            },
+        ),
+        use_system_prompt=True,
+    )
+    prompt_generator.load()
+    return prompt_generator
+
+
 def get_magpie_generator(num_turns, num_rows, system_prompt, is_sample):
     input_mappings = _get_output_mappings(num_turns)
     output_mappings = input_mappings.copy()
@@ -260,21 +278,3 @@ def get_response_generator(num_turns, system_prompt, is_sample):
         )
     response_generator.load()
     return response_generator
-
-
-def get_prompt_generator():
-    prompt_generator = TextGeneration(
-        llm=InferenceEndpointsLLM(
-            api_key=_get_next_api_key(),
-            model_id=MODEL,
-            tokenizer_id=MODEL,
-            generation_kwargs={
-                "temperature": 0.8,
-                "max_new_tokens": 2048,
-                "do_sample": True,
-            },
-        ),
-        use_system_prompt=True,
-    )
-    prompt_generator.load()
-    return prompt_generator

src/distilabel_dataset_generator/pipelines/textcat.py

@@ -148,6 +148,24 @@ with Pipeline(name="textcat") as pipeline:
     )
 
 
+def get_prompt_generator(temperature):
+    prompt_generator = TextGeneration(
+        llm=InferenceEndpointsLLM(
+            api_key=_get_next_api_key(),
+            model_id=MODEL,
+            tokenizer_id=MODEL,
+            generation_kwargs={
+                "temperature": temperature,
+                "max_new_tokens": 2048,
+                "do_sample": True,
+            },
+        ),
+        use_system_prompt=True,
+    )
+    prompt_generator.load()
+    return prompt_generator
+
+
 def get_textcat_generator(difficulty, clarity, is_sample):
     textcat_generator = GenerateTextClassificationData(
         llm=InferenceEndpointsLLM(
@@ -188,21 +206,3 @@ def get_labeller_generator(system_prompt, labels, num_labels):
     )
     labeller_generator.load()
     return labeller_generator
-
-
-def get_prompt_generator():
-    prompt_generator = TextGeneration(
-        llm=InferenceEndpointsLLM(
-            api_key=_get_next_api_key(),
-            model_id=MODEL,
-            tokenizer_id=MODEL,
-            generation_kwargs={
-                "temperature": 0.8,
-                "max_new_tokens": 2048,
-                "do_sample": True,
-            },
-        ),
-        use_system_prompt=True,
-    )
-    prompt_generator.load()
-    return prompt_generator