Upload folder using huggingface_hub

config.json

@@ -0,0 +1,15 @@
+{
+    "model_type": "gpt2",
+    "pipeline_tag": "text-generation",
+    "auto_map": {
+        "pipeline": "tiny_chained_tossup.QBChainedPipeline"
+    },
+    "custom_pipelines": {
+        "quizbowl-tossup": {
+            "impl": "tiny_chained_tossup.QBChainedPipeline",
+            "pt": [],
+            "tf": [],
+            "type": "text"
+        }
+    }
+}

stage1/config.json

@@ -0,0 +1,38 @@
+{
+  "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": "4.52.2",
+  "use_cache": true,
+  "vocab_size": 50257
+}

stage1/generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 50256,
+  "eos_token_id": 50256,
+  "transformers_version": "4.52.2"
+}

stage1/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7d00560d8910fbed77ffad4065dee5011c41ba401b1064e749c498ba9e20373
+size 497774208

stage1/special_tokens_map.json

@@ -0,0 +1,5 @@
+{
+  "bos_token": "<|endoftext|>",
+  "eos_token": "<|endoftext|>",
+  "unk_token": "<|endoftext|>"
+}

stage1/tokenizer_config.json

@@ -0,0 +1,20 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "50256": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<|endoftext|>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "extra_special_tokens": {},
+  "model_max_length": 1024,
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>"
+}

stage2/config.json

@@ -0,0 +1,45 @@
+{
+  "_num_labels": 1,
+  "activation_function": "gelu_new",
+  "architectures": [
+    "GPT2LMHeadModel"
+  ],
+  "attn_pdrop": 0.1,
+  "bos_token_id": 50256,
+  "embd_pdrop": 0.1,
+  "eos_token_id": 50256,
+  "id2label": {
+    "0": "LABEL_0"
+  },
+  "initializer_range": 0.02,
+  "label2id": {
+    "LABEL_0": 0
+  },
+  "layer_norm_epsilon": 1e-05,
+  "model_type": "gpt2",
+  "n_ctx": 1024,
+  "n_embd": 768,
+  "n_head": 12,
+  "n_inner": null,
+  "n_layer": 6,
+  "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": "4.52.2",
+  "use_cache": true,
+  "vocab_size": 50257
+}

stage2/generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 50256,
+  "eos_token_id": 50256,
+  "transformers_version": "4.52.2"
+}

stage2/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:721fa38acb2ba5d1ebdeed4bad3dcb32c0beeb2b924ecf544fffaa3b0bb65a66
+size 327657928

stage2/special_tokens_map.json

@@ -0,0 +1,5 @@
+{
+  "bos_token": "<|endoftext|>",
+  "eos_token": "<|endoftext|>",
+  "unk_token": "<|endoftext|>"
+}

stage2/tokenizer_config.json

@@ -0,0 +1,20 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "50256": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<|endoftext|>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "extra_special_tokens": {},
+  "model_max_length": 1024,
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>"
+}

tiny_chained_tossup.py

@@ -0,0 +1,80 @@
+"""
+Chained pipeline:
+  • Stage-A: classify hypothesis vs. premise → label distribution (Y)
+  • Stage-B: consume Y (+ original text) → final dict {guess, confidence, explanation}
+The outer task id is "qpe-tossup" (or "qpe-bonus", pick one).
+"""
+
+# %%s
+import os
+
+from dotenv import load_dotenv
+
+load_dotenv(os.path.expanduser("~/.zshenv"))
+from pathlib import Path
+
+import torch
+import torch.nn.functional as F
+from transformers import (
+    AutoModelForSequenceClassification,
+    AutoTokenizer,
+    Pipeline,
+    pipeline,
+)
+
+
+class QBChainedPipeline(Pipeline):
+    def __init__(self, pipeA, pipeB, **kwargs):
+        if pipeA.task != "text-generation" or pipeB.task != "text-generation":
+            raise ValueError(
+                f"Both stages must be text-generation pipelines, but got {pipeA.task} and {pipeB.task}"
+            )
+        self.pipeA = pipeA
+        self.pipeB = pipeB
+        super().__init__(
+            model=None, tokenizer=None, task="quizbowl-tossup", **kwargs
+        )  # we call sub-pipelines manually
+
+    # ---- Standard Pipeline API overrides ----------------------------------
+
+    def _sanitize_parameters(self, **pipeline_kwargs):
+        """
+        Forward every keyword argument to the inner generators unchanged.
+        """
+        return pipeline_kwargs, {}, {}
+
+    def preprocess(self, inputs, **kwargs):
+        """
+        No special preprocessing—just forward the raw input string.
+        """
+        return inputs
+
+    def _forward(self, inputs, *args, **kw):
+        """
+        inputs: {"question_text": str}
+        returns: {"answer": str, "confidence": float, "explanation": str}
+        """
+        question_text = inputs["question_text"]
+        # -------- Stage A --------
+        promptA = f"Answer the quiz question revealed so far:\n Question:{question_text}\nAnswer:"
+        outA = self.pipeA(promptA)[0]["generated_text"]
+        answer = outA.removeprefix(promptA).strip()
+
+        # -------- Stage B --------
+        promptB = f"Given the question and the answer, output the confidence score between 0 and 1:\n Question:{question_text}\nAnswer: {answer}\nConfidence:"
+        outB = self.pipeB(promptB)[0]["generated_text"]
+        confidence = float(outB.removeprefix(promptB).strip())
+
+        return {"answer": answer, "confidence": confidence}
+
+    def postprocess(self, model_outputs):
+        """
+        Hugging Face pipelines expect structured output.
+        """
+        return {
+            **model_outputs,
+            "explanation": f"Confidence score for the answer '{model_outputs['answer']}' is {model_outputs['confidence']}.",
+        }
+
+
+# %%