mriusero commited on
Commit
7f568ce
·
1 Parent(s): b6f3c79

large to medium llm

Browse files
Files changed (3) hide show
  1. README.md +2 -2
  2. app.py +1 -1
  3. src/agent/stream.py +0 -18
README.md CHANGED
@@ -35,8 +35,8 @@ This is a demo of an AI agent designed to assist industries and service provider
35
 
36
  ### Design
37
 
38
- * The agent is implemented using **Mistral AI** via the `mistral-large-2411` LLM.
39
- * Its capabilities have been enhanced with a chain-of-thought reasoning process, allowing it to think, act, observe, and respond effectively to user queries.
40
  * The agent is presented through a **Gradio interface**, which is well-suited for both real-time visualization and LLM interaction.
41
 
42
  ### Purposes
 
35
 
36
  ### Design
37
 
38
+ * The agent is implemented using **Mistral AI** via the **mistral-medium-2505** LLM.
39
+ * Its capabilities have been enhanced with a **chain-of-thought** reasoning process, allowing it to think, act, observe, and respond effectively to user queries.
40
  * The agent is presented through a **Gradio interface**, which is well-suited for both real-time visualization and LLM interaction.
41
 
42
  ### Purposes
app.py CHANGED
@@ -48,7 +48,7 @@ This is a demo of an AI agent designed to assist industries and service provider
48
  """
49
  ## Design
50
 
51
- The agent is implemented using **Mistral AI** via the `mistral-large-2411` LLM. Its capabilities have been enhanced with a chain-of-thought reasoning process, allowing it to `think`, `act`, `observe`, and `respond` effectively to user queries. The agent is presented through a **Gradio interface**, which is well-suited for both real-time visualization and LLM interaction.
52
 
53
  [See video overview](https://drive.google.com/file/d/1Bv1uF3-4EeR1HePSafZN1yzInr7YcQXZ/view?usp=share_link)
54
 
 
48
  """
49
  ## Design
50
 
51
+ The agent is implemented using **Mistral AI** via the `mistral-medium-2505` LLM. Its capabilities have been enhanced with a chain-of-thought reasoning process, allowing it to `think`, `act`, `observe`, and `respond` effectively to user queries. The agent is presented through a **Gradio interface**, which is well-suited for both real-time visualization and LLM interaction.
52
 
53
  [See video overview](https://drive.google.com/file/d/1Bv1uF3-4EeR1HePSafZN1yzInr7YcQXZ/view?usp=share_link)
54
 
src/agent/stream.py CHANGED
@@ -92,24 +92,6 @@ async def respond(message, history=None, state=None):
92
  if current_phase == "think":
93
  history[-1] = ChatMessage(role="assistant", content=buffer, metadata={"title": "Thinking...", "status": "pending", "id": state['cycle']})
94
 
95
- #elif current_phase == "act":
96
- #parent_message = next((msg for msg in history if msg.metadata.get("id") == state['cycle']), None)
97
- #if parent_message:
98
- # parent_message.content += "\n\n" + buffer
99
- # parent_message.metadata["title"] = "Acting..."
100
- #else:
101
- # history[-1] = ChatMessage(role="assistant", content=buffer, metadata={"title": "Acting...", "status": "pending", "id": state['cycle']+1, 'parent_id': state["cycle"]})
102
-
103
- #elif current_phase == "observe":
104
- # parent_message = next((msg for msg in history if msg.metadata.get("id") == state['cycle']), None)
105
- # if parent_message:
106
- # parent_message.content += "\n\n" + buffer
107
- # parent_message.metadata["title"] = "Acting..."
108
- # else:
109
- # history[-1] = ChatMessage(role="assistant", content=buffer, metadata={"title": "Observing...", "status": "pending", "id": state['cycle']+2, 'parent_id': state["cycle"]})
110
- #
111
- #yield history
112
-
113
  if current_phase == "final":
114
  delta_content = delta.content or ""
115
  final_full += delta_content
 
92
  if current_phase == "think":
93
  history[-1] = ChatMessage(role="assistant", content=buffer, metadata={"title": "Thinking...", "status": "pending", "id": state['cycle']})
94
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
95
  if current_phase == "final":
96
  delta_content = delta.content or ""
97
  final_full += delta_content