add all files

README.md

@@ -1,12 +1,5 @@
----
-title: KhaanaGPT
-emoji: 🦀
-colorFrom: indigo
-colorTo: gray
-sdk: gradio
-sdk_version: 3.18.0
-app_file: app.py
-pinned: false
----
+# खानाGPT
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+Just a basic app to generate Indian recipes from listed ingredients. It's good, not great, it might add more ingredients or ignore some existing ingredients.
+
+This was my first attempt at fine-tuning causalLM models. I used `GPT-2 Small` for this and it was fun, learnt a lot and the outputs are as expected.

app.py

@@ -0,0 +1,59 @@
+from transformers import pipeline
+import gradio as gr
+
+import warnings
+warnings.simplefilter('ignore')
+
+model_path = './khaanaGPT'
+
+contrastive_search_config = dict(
+    penalty_alpha = 0.5,
+    top_k = 5,
+    max_new_tokens = 512,
+    pad_token_id = 50259
+)
+
+model = pipeline('text-generation',model=model_path)
+
+def create_prompt(ingredients):
+    ingredients = ','.join([x.strip() for x in ingredients.split(',')])
+    ingredients = ingredients.strip().replace(',','\n').lower()
+    s = f"<|startoftext|>Ingredients:\n{ingredients}\n\nInstructions:\n"
+    return s
+
+def generate(prompt):
+    recipe = model(prompt,**contrastive_search_config)[0]['generated_text']
+    recipe = recipe.replace('<|startoftext|>','')
+    return recipe
+
+def wrapper(ingredients):
+    prompt = create_prompt(ingredients)
+    recipe = generate(prompt)
+    return recipe
+
+intro_html = """
+<center><h1>खानाGPT</h1></center>
+<center>
+<p>it's not perfect, may ± ingredients. The recipes are coherent, 
+but the main purpose of this project was to understand fine-tuning a causalLM like GPT-2.
+This model was fine-tuned on GPT-2 Small.</p>
+</center>
+"""
+
+with gr.Blocks() as demo:
+    gr.HTML(intro_html)
+
+    ingredients = gr.Textbox(label="ingredients",
+    placeholder='separate the ingredients with a comma.')
+
+    output = gr.Textbox(label="recipe",lines=15,)
+    greet_btn = gr.Button("Create a recipe!")
+
+    gr.Examples(['yellow dal, turmeric, green peas, tomatoes',
+                'chicken, soy sauce, tomato sauce, vinegar'],
+                inputs=ingredients
+            )
+
+    greet_btn.click(fn=wrapper, inputs=ingredients, outputs=output)
+
+demo.launch()

khaanaGPT/added_tokens.json

@@ -0,0 +1,5 @@
+{
+  "<|pad|>": 50259,
+  "<|startoftext|>": 50257,
+  "<|unknown|>": 50258
+}

khaanaGPT/config.json

@@ -0,0 +1,39 @@
+{
+  "_name_or_path": "gpt2",
+  "activation_function": "gelu_new",
+  "architectures": [
+    "GPT2LMHeadModel"
+  ],
+  "attn_pdrop": 0.1,
+  "bos_token_id": 50256,
+  "embd_pdrop": 0.1,
+  "eos_token_id": 50256,
+  "initializer_range": 0.02,
+  "layer_norm_epsilon": 1e-05,
+  "model_type": "gpt2",
+  "n_ctx": 1024,
+  "n_embd": 768,
+  "n_head": 12,
+  "n_inner": null,
+  "n_layer": 12,
+  "n_positions": 1024,
+  "reorder_and_upcast_attn": false,
+  "resid_pdrop": 0.1,
+  "scale_attn_by_inverse_layer_idx": false,
+  "scale_attn_weights": true,
+  "summary_activation": null,
+  "summary_first_dropout": 0.1,
+  "summary_proj_to_labels": true,
+  "summary_type": "cls_index",
+  "summary_use_proj": true,
+  "task_specific_params": {
+    "text-generation": {
+      "do_sample": true,
+      "max_length": 50
+    }
+  },
+  "torch_dtype": "float32",
+  "transformers_version": "4.20.1",
+  "use_cache": true,
+  "vocab_size": 50260
+}

khaanaGPT/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3565459822f2fe308ed7e0ac736366ef029c2f64b69379ab11306f805f3157ea
+size 510405737

khaanaGPT/special_tokens_map.json

@@ -0,0 +1,6 @@
+{
+  "bos_token": "<|startoftext|>",
+  "eos_token": "<|endoftext|>",
+  "pad_token": "<|pad|>",
+  "unk_token": "<|unknown|>"
+}

khaanaGPT/tokenizer_config.json

@@ -0,0 +1,11 @@
+{
+  "add_prefix_space": false,
+  "bos_token": "<|startoftext|>",
+  "eos_token": "<|endoftext|>",
+  "model_max_length": 1024,
+  "name_or_path": "gpt2",
+  "pad_token": "<|pad|>",
+  "special_tokens_map_file": null,
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|unknown|>"
+}

khaanaGPT/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad32707bc4b19e7bc4f7e6089539caab91af3f0984c90926718351e222e31e12
+size 3247

pyproject.toml

@@ -0,0 +1,17 @@
+[tool.poetry]
+name = "khaanagpt"
+version = "0.1.0"
+description = ""
+authors = ["shreydan <iamshreydan@gmail.com>"]
+readme = "README.md"
+
+[tool.poetry.dependencies]
+python = "^3.10"
+transformers = "^4.26.1"
+torch = "^1.13.1"
+gradio = "^3.18.0"
+
+
+[build-system]
+requires = ["poetry-core"]
+build-backend = "poetry.core.masonry.api"

requirements.txt

@@ -0,0 +1,3 @@
+transformers
+torch
+gradio