From 644c3ce56a0e976db273c4307ce39e965bfe274d Mon Sep 17 00:00:00 2001
From: Gaurav Shukla <gaurav@nod-labs.com>
Date: Fri, 23 Jun 2023 20:06:44 +0530
Subject: [PATCH] [vicuna] Add streaming of tokens

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
---
 apps/language_models/src/pipelines/vicuna_pipeline.py | 3 +++
 apps/stable_diffusion/web/ui/stablelm_ui.py           | 8 +++-----
 2 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/apps/language_models/src/pipelines/vicuna_pipeline.py b/apps/language_models/src/pipelines/vicuna_pipeline.py
index 3c9c123c67..0e418c65f0 100644
--- a/apps/language_models/src/pipelines/vicuna_pipeline.py
+++ b/apps/language_models/src/pipelines/vicuna_pipeline.py
@@ -461,6 +461,7 @@ def generate(self, prompt, cli=False):
         logits = generated_token_op["logits"]
         pkv = generated_token_op["pkv"]
         detok = generated_token_op["detok"]
+        yield detok
 
         res.append(detok)
         res_tokens.append(token)
@@ -502,6 +503,8 @@ def generate(self, prompt, cli=False):
                 res.append(detok)
                 if cli:
                     print(f"{detok}", end=" ", flush=True)
+            yield detok
+
         if self.device == "cuda":
             del sec_vic, pkv, logits
             torch.cuda.empty_cache()
diff --git a/apps/stable_diffusion/web/ui/stablelm_ui.py b/apps/stable_diffusion/web/ui/stablelm_ui.py
index 0e5cf4092d..84d5921fcc 100644
--- a/apps/stable_diffusion/web/ui/stablelm_ui.py
+++ b/apps/stable_diffusion/web/ui/stablelm_ui.py
@@ -65,16 +65,14 @@ def chat(curr_system_message, history, model, device, precision):
         )
         prompt = messages.strip()
         print("prompt = ", prompt)
-        sentence = vicuna_model.generate(prompt)
 
         partial_text = ""
-        for new_text in sentence.split(" "):
+        for new_text in vicuna_model.generate(prompt):
             # print(new_text)
             partial_text += new_text + " "
             history[-1][1] = partial_text
             # Yield an empty string to cleanup the message textbox and the updated conversation history
             yield history
-        history[-1][1] = sentence
         return history
 
     # else Model is StableLM
@@ -124,13 +122,13 @@ def chat(curr_system_message, history, model, device, precision):
                 "TheBloke/vicuna-7B-1.1-HF",
             ],
         )
-        supported_devices = available_devices
+        supported_devices = available_devices + ["AMD-AIE"]
         enabled = len(supported_devices) > 0
         device = gr.Dropdown(
             label="Device",
             value=supported_devices[0]
             if enabled
-            else "Only CUDA Supported for now",
+            else "No devices supported for now",
             choices=supported_devices,
             interactive=enabled,
         )