Duplicate from EuroPython2022/Scratchpad-w-BLOOM

Co-authored-by: Muhtasham Oblokulov <[email protected]>

.gitattributes

@@ -0,0 +1,27 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,13 @@
+---
+title: Length generalization via BLOOM
+emoji: 📝
+colorFrom: indigo
+colorTo: red
+sdk: gradio
+sdk_version: 3.0.24
+app_file: app.py
+pinned: false
+duplicated_from: EuroPython2022/Scratchpad-w-BLOOM
+---
+
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -0,0 +1,63 @@
+import gradio as gr
+import requests
+import os
+
+##Bloom
+API_URL = "https://api-inference.huggingface.co/models/bigscience/bloom"
+HF_TOKEN = os.environ["HF_TOKEN"]
+headers = {"Authorization": f"Bearer {HF_TOKEN}"}
+
+def text_generate(prompt):
+  print(f"Prompt is :{prompt}")
+  p =  prompt + " Solution: " 
+  print(f"Final prompt is : {p}")
+  json_ = {"inputs": p,
+            "parameters":
+            {
+            "top_p": 0.9,
+          "temperature": 1.1,
+          "max_new_tokens": 250,
+          "return_full_text": True
+          }, "options": 
+              {
+              "use_cache": True,
+              "wait_for_model":True
+              },}
+  response = requests.post(API_URL, headers=headers, json=json_)
+  print(f"Response  is : {response}")
+  output = response.json()
+  print(f"output is : {output}")
+  output_tmp = output[0]['generated_text']
+  print(f"output_tmp is: {output_tmp}")
+  solution = output_tmp.split("\nQ:")[0]
+  print(f"Final response after splits is: {solution}") 
+  return solution 
+
+demo = gr.Blocks()
+
+with demo:
+    gr.Markdown("<h1><center>Length generalization (LG) With BLOOM🌸 </center></h1>")
+    gr.Markdown(
+            """
+            We will examine large language models ability to extrapolate to longer problems! \n
+            Length generalization (LG) is important: Often, long examples are rare and intrinsically more difficult, yet are the ones we care more about.  \n
+            Recent paper [Exploring Length Generalization in Large Language Models](https://arxiv.org/pdf/2207.04901) found that using few-shot  [scratchpad](https://arxiv.org/abs/2112.00114), a combo behind many strong LLM results (eg. #Minerva ) \n
+            leads to **substantial improvements in length generalization!** \n
+            In-context learning enables variable length pattern matching, producing solutions of correct lengths. \n
+            This space is an attempt at inspecting this LLM behavior/capability in the new HuggingFace BigScienceW [Bloom](https://huggingface.co/bigscience/bloom) model. \n
+            This Space is created by [Muhtasham Oblokulov](https://twitter.com/muhtasham9) for EuroPython 2022 Demo. \n
+            This Space is work in progress, BLOOM doesn't support inference on long sequencess so you may try with shorter sequences. \n
+            """
+            )
+    with gr.Row(): 
+        input_prompt = gr.Textbox(value="Q:The coin is heads up.(1) Then Austin flips. Is the coin still heads up? Solution: Coin is initially heads up. (1) After Austin flips, coin turns to heads. Q: The coin is heads up. (2) Then Austin doesn't flip. (1) Then Kara flips. Is the coin still heads up?",
+        label="Enter your examples zero-shot (few-shot is not supported due to API limit) followed by Query :")
+        generated_txt = gr.Textbox(lines=10, label="Generated Solution:")
+
+    b1 = gr.Button("Generate Text")
+    b1.click(text_generate,inputs=[input_prompt], outputs=[generated_txt])
+    
+    with gr.Row(): 
+            gr.Markdown("![visitor badge](https://visitor-badge.glitch.me/badge?page_id=europython2022_scratchpad-w-bloom)")
+
+demo.launch(enable_queue=True, debug=True)