Instructions to use org2ai/Wald-4B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use org2ai/Wald-4B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="org2ai/Wald-4B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("org2ai/Wald-4B")
model = AutoModelForCausalLM.from_pretrained("org2ai/Wald-4B", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Laya

How to use org2ai/Wald-4B with Laya:

# No code snippets available yet for this library.

# To use this model, check the repository files and the library's documentation.

# Want to help? PRs adding snippets are welcome at:
# https://github.com/huggingface/huggingface.js

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use org2ai/Wald-4B with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf org2ai/Wald-4B:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf org2ai/Wald-4B:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf org2ai/Wald-4B:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf org2ai/Wald-4B:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf org2ai/Wald-4B:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf org2ai/Wald-4B:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf org2ai/Wald-4B:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf org2ai/Wald-4B:Q4_K_M

Use Docker

docker model run hf.co/org2ai/Wald-4B:Q4_K_M

LM Studio
Jan

vLLM

How to use org2ai/Wald-4B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "org2ai/Wald-4B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "org2ai/Wald-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/org2ai/Wald-4B:Q4_K_M

SGLang

How to use org2ai/Wald-4B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "org2ai/Wald-4B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "org2ai/Wald-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "org2ai/Wald-4B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "org2ai/Wald-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use org2ai/Wald-4B with Ollama:
```
ollama run hf.co/org2ai/Wald-4B:Q4_K_M
```
Unsloth Desktop

How to use org2ai/Wald-4B with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf org2ai/Wald-4B:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "org2ai/Wald-4B:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use org2ai/Wald-4B with Docker Model Runner:
```
docker model run hf.co/org2ai/Wald-4B:Q4_K_M
```

Lemonade

How to use org2ai/Wald-4B with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull org2ai/Wald-4B:Q4_K_M

Run and chat with the model

lemonade run user.Wald-4B-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use org2ai/Wald-4B with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf org2ai/Wald-4B:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default org2ai/Wald-4B:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use org2ai/Wald-4B with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf org2ai/Wald-4B:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "org2ai/Wald-4B:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Wald-4B / contamination /trained-on-di-ids.json

Harry19081

Wald-4B v1.0

78ebf43 7 days ago

Raw History Blame Contribute Delete

134 kB

	{
	"schema": "wald/di-trained-on/1",
	"what": "Decision Index 0.2 / 0.2.1 item ids present in the training data of the submitted checkpoint's lineage (strict match: the whole state in one training record, plus >= 50 % of option text where options are the item's content). Count every one of them as trained on.",
	"total": 363,
	"per_benchmark": [
	{
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"items": 67
	},
	{
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"items": 49
	},
	{
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"items": 44
	},
	{
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"items": 170
	},
	{
	"benchmark_id": 12,
	"benchmark": "ANLI",
	"scored_in_0_2_1": "scored",
	"items": 3
	},
	{
	"benchmark_id": 24,
	"benchmark": "MMLU",
	"scored_in_0_2_1": "display only",
	"items": 6
	},
	{
	"benchmark_id": 26,
	"benchmark": "ARC-Easy",
	"scored_in_0_2_1": "display only",
	"items": 7
	},
	{
	"benchmark_id": 27,
	"benchmark": "ARC-Challenge",
	"scored_in_0_2_1": "display only",
	"items": 4
	},
	{
	"benchmark_id": 28,
	"benchmark": "WinoGrande",
	"scored_in_0_2_1": "scored",
	"items": 2
	},
	{
	"benchmark_id": 33,
	"benchmark": "SATA-Bench",
	"scored_in_0_2_1": "scored",
	"items": 4
	},
	{
	"benchmark_id": 36,
	"benchmark": "BRIGHT",
	"scored_in_0_2_1": "scored",
	"items": 1
	},
	{
	"benchmark_id": 57,
	"benchmark": "MMLU-Pro",
	"scored_in_0_2_1": "scored",
	"items": 6
	}
	],
	"scans": {
	"stage 1 (full-parameter decision training corpus)": {
	"strict_hits_in_trained_files": 177,
	"blocklist_sha256": "a9dc4133464334273a6326c125744beaf6e44d45eba1d25fae5b61225e1ad4ba",
	"suite_fingerprints_sha256": "df92bd3ceed2f57df7d25e219a22e18521097db625403b5c3de90226b809b2a5",
	"files": {
	"train-00000.jsonl": {
	"lines": 92013,
	"sha256": "ed48083526b4b063ee8be3bc921b018ef1ec63d2c308e03d8ee57574c292431b",
	"strict": 36
	},
	"train-00001.jsonl": {
	"lines": 92531,
	"sha256": "ac12385f728387d46cb97a9323c42be6e5d5f32ef83f0a257efbf9f8db980bc4",
	"strict": 44
	},
	"train-00002.jsonl": {
	"lines": 92089,
	"sha256": "10bbbe85daaafef4d575abc3d12b3738d1484e93caf64129ffbba2b98413f86a",
	"strict": 44
	},
	"train-00003.jsonl": {
	"lines": 92440,
	"sha256": "3a883ad81670c1b51dd583e1eaf346e4abe4336cfcd59a15dd1227f831e33052",
	"strict": 53
	},
	"train-00004.jsonl": {
	"lines": 2900,
	"sha256": "8b3e1b34ab30c8746f8858047aa31dd39f42a4bdaebe46786e19e9762aa02e9a",
	"strict": 0
	}
	}
	},
	"stage 2 (LoRA refinement data)": {
	"strict_hits_in_trained_files": 42,
	"blocklist_sha256": "a9dc4133464334273a6326c125744beaf6e44d45eba1d25fae5b61225e1ad4ba",
	"suite_fingerprints_sha256": "df92bd3ceed2f57df7d25e219a22e18521097db625403b5c3de90226b809b2a5",
	"files": {
	"train-f3base-00000.jsonl": {
	"lines": 18441,
	"sha256": "9a882f66a815d27bf31264fc19a9680e04bf59340d0b275d87d017925a46a870",
	"strict": 36
	},
	"train-replay-00000.jsonl": {
	"lines": 6746,
	"sha256": "b236f01529a76d45dbcf5a049b629fc8c9da2f30eab6e14b615360b248dc608c",
	"strict": 6
	},
	"train-teacher-00000.jsonl": {
	"lines": 63,
	"sha256": "ae4ce672535b2fa9a11af0df8f2866e61a2cdbe80042d1f5607778d60e0af98c",
	"strict": 0
	}
	}
	},
	"stage 3 (short-thought distillation data)": {
	"strict_hits_in_trained_files": 158,
	"blocklist_sha256": "a9dc4133464334273a6326c125744beaf6e44d45eba1d25fae5b61225e1ad4ba",
	"suite_fingerprints_sha256": "df92bd3ceed2f57df7d25e219a22e18521097db625403b5c3de90226b809b2a5",
	"files": {
	"train-final.jsonl": {
	"lines": 33361,
	"sha256": "ecaaf022dd5376e6823d2c5f8acd9602e3cc76a06fb582a72cb3287edb311d9e",
	"strict": 79
	},
	"train.jsonl": {
	"lines": 33361,
	"sha256": "846caf99110bff0d8c26e35c43aee35da72d44b33b26599b1fb53d9dee762a0d",
	"strict": 79
	}
	}
	}
	},
	"stage_4_targeted_lora": "stage-4 data (Home-appliance generator rows; iSarcasmEval / API-Bank / ContractNLI / VAST / NLI4CT / ACOS / RAGTruth train splits; replay) was checked against the full blocklist and the sample rows before training: 0 hits after filtering",
	"items": [
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:craft-math-algebra:craft_Math_algebra_query_170:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_109:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_128:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_12:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_13:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_191:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_194:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_194:chunk1",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_213:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_217:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_232:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_232:chunk1",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_234:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_238:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_241:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_257:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_299:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_307:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_323:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_340:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_343:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_343:chunk1",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_343:chunk2",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_34:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_357:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_358:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_363:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_378:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_37:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_386:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_392:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_395:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_406:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_406:chunk1",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_420:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_422:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_439:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_45:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_516:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_516:chunk1",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_570:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_575:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_57:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_57:chunk1",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_580:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_59:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_601:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_603:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_632:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_65:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_667:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_687:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_705:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_705:chunk1",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_734:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_742:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_766:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_781:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_796:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_85:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_86:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_888:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_905:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_90:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_91:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_91:chunk1",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "2:ToolRet-retrieval:ToolRet-retrieval:toolace:toolACE_query_980:chunk0",
	"benchmark_id": 2,
	"benchmark": "ToolRet",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: ToolRet aggregates user queries of public tool-calling train sets (ToolACE, Glaive) that are in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1197",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1198",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1246",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1289",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1322",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1332",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1342",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1432",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1437",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1474",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1573",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1616",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1687",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1735",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1817",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1821",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1936",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1946",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:1993",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2083",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2139",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2145",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2149",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2254",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2298",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2331",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2381",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2436",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2441",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2466",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2476",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:250",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:256",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2601",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2605",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2755",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:2821",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:3012",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:3070",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:3074",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:347",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:391",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:450",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:460",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:510",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:888",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:890",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:891",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "4:BANKING77:BANKING77:test:893",
	"benchmark_id": 4,
	"benchmark": "BANKING77",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the benchmark's public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:1055",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:1126",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:1132",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:1215",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:1233",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:1252",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:1332",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:1333",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:1487",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:1505",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:1518",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:1809",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:1922",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:2234",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:2236",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:2243",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:2246",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:2372",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:2446",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:2658",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:274",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:2752",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3316",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3321",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3629",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3630",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3631",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3637",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3643",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3644",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3645",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3652",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3658",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3667",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3679",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3705",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3808",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:3865",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:4281",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:4457",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:659",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:671",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:70",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "5:CLINC150+OOS:CLINC150+OOS:test:930",
	"benchmark_id": 5,
	"benchmark": "CLINC150+OOS",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence as the test item",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:10002",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:10008",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:10026",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:10027",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:10060",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:10071",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:10180",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:10345",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:10350",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:10548",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:3624",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:3645",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:3661",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:3795",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:4005",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:4168",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:4223",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:4296",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:4367",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:4388",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:4422",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:4434",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:4438",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:4497",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:4558",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:4769",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:4807",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:4829",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:5056",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:5116",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:5153",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:5192",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:5208",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:5227",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:5240",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:5244",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:5326",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:5451",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:5480",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:5501",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:5597",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:6265",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:6566",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:6659",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:6699",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:6733",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:6743",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:6795",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:6872",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:6916",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:7172",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:7209",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:7270",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:7524",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:7566",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:7705",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:7964",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8017",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8023",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8079",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8130",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8131",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8302",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8421",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8525",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8634",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8648",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8691",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8718",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8766",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8807",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8830",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8913",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:8943",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:9255",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:9256",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:9292",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:9363",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:9411",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:9431",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:9439",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:9484",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:9620",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:9651",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-0shot:RouterBench-0shot:9801",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:10016",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:10022",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:10040",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:10041",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:10074",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:10085",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:10194",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:10359",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:10364",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:10562",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:3638",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:3659",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:3675",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:3809",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:4019",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:4182",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:4237",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:4310",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:4381",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:4402",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:4436",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:4448",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:4452",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:4511",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:4572",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:4783",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:4821",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:4843",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:5070",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:5130",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:5167",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:5206",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:5222",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:5241",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:5254",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:5258",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:5340",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:5465",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:5494",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:5515",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:5611",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:6279",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:6580",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:6673",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:6713",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:6747",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:6757",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:6809",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:6886",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:6930",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:7186",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:7223",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:7284",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:7538",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:7580",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:7719",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:7978",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8031",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8037",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8093",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8144",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8145",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8316",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8435",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8539",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8648",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8662",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8705",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8732",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8780",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8821",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8844",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8927",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:8957",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:9269",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:9270",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:9306",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:9377",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:9425",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:9445",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:9453",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:9498",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:9634",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:9665",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "6:RouterBench-5shot:RouterBench-5shot:9815",
	"benchmark_id": 6,
	"benchmark": "RouterBench",
	"scored_in_0_2_1": "not scored in 0.2.1",
	"why": "shared upstream source: RouterBench prompts embed GSM8K / MMLU questions that are in our math and QA data",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "12:ANLI:ANLI:test_r2:4bb4236d-e277-443c-9a83-05aa3e1f36ab",
	"benchmark_id": 12,
	"benchmark": "ANLI",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream text: the premise text also appears in another public dataset in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "12:ANLI:ANLI:test_r2:5c674fb6-d204-46ff-8ab1-19b483d97f14",
	"benchmark_id": 12,
	"benchmark": "ANLI",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream text: the premise text also appears in another public dataset in our data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "12:ANLI:ANLI:test_r3:af91dd15-20c2-4428-9bdf-8eb8eb1ea180",
	"benchmark_id": 12,
	"benchmark": "ANLI",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream text: the premise text also appears in another public dataset in our data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "24:MMLU:MMLU:test:4250",
	"benchmark_id": 24,
	"benchmark": "MMLU",
	"scored_in_0_2_1": "display only",
	"why": "question also present in MMLU auxiliary_train / other public QA sets",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "24:MMLU:MMLU:test:4261",
	"benchmark_id": 24,
	"benchmark": "MMLU",
	"scored_in_0_2_1": "display only",
	"why": "question also present in MMLU auxiliary_train / other public QA sets",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "24:MMLU:MMLU:test:4274",
	"benchmark_id": 24,
	"benchmark": "MMLU",
	"scored_in_0_2_1": "display only",
	"why": "question also present in MMLU auxiliary_train / other public QA sets",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "24:MMLU:MMLU:test:4281",
	"benchmark_id": 24,
	"benchmark": "MMLU",
	"scored_in_0_2_1": "display only",
	"why": "question also present in MMLU auxiliary_train / other public QA sets",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "24:MMLU:MMLU:test:4419",
	"benchmark_id": 24,
	"benchmark": "MMLU",
	"scored_in_0_2_1": "display only",
	"why": "question also present in MMLU auxiliary_train / other public QA sets",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "24:MMLU:MMLU:test:7920",
	"benchmark_id": 24,
	"benchmark": "MMLU",
	"scored_in_0_2_1": "display only",
	"why": "question also present in MMLU auxiliary_train / other public QA sets",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "26:ARC-Easy:ARC-Easy:test:CSZ20770",
	"benchmark_id": 26,
	"benchmark": "ARC-Easy",
	"scored_in_0_2_1": "display only",
	"why": "train-split duplicate (ARC train / MMLU auxiliary_train), the board's known duplicates",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "26:ARC-Easy:ARC-Easy:test:CSZ30768",
	"benchmark_id": 26,
	"benchmark": "ARC-Easy",
	"scored_in_0_2_1": "display only",
	"why": "train-split duplicate (ARC train / MMLU auxiliary_train), the board's known duplicates",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "26:ARC-Easy:ARC-Easy:test:LEAP__4_10224",
	"benchmark_id": 26,
	"benchmark": "ARC-Easy",
	"scored_in_0_2_1": "display only",
	"why": "train-split duplicate (ARC train / MMLU auxiliary_train), the board's known duplicates",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "26:ARC-Easy:ARC-Easy:test:MDSA_2013_8_7",
	"benchmark_id": 26,
	"benchmark": "ARC-Easy",
	"scored_in_0_2_1": "display only",
	"why": "train-split duplicate (ARC train / MMLU auxiliary_train), the board's known duplicates",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "26:ARC-Easy:ARC-Easy:test:Mercury_7172813",
	"benchmark_id": 26,
	"benchmark": "ARC-Easy",
	"scored_in_0_2_1": "display only",
	"why": "train-split duplicate (ARC train / MMLU auxiliary_train), the board's known duplicates",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "26:ARC-Easy:ARC-Easy:test:NYSEDREGENTS_2012_4_20",
	"benchmark_id": 26,
	"benchmark": "ARC-Easy",
	"scored_in_0_2_1": "display only",
	"why": "train-split duplicate (ARC train / MMLU auxiliary_train), the board's known duplicates",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "26:ARC-Easy:ARC-Easy:test:NYSEDREGENTS_2012_8_18",
	"benchmark_id": 26,
	"benchmark": "ARC-Easy",
	"scored_in_0_2_1": "display only",
	"why": "train-split duplicate (ARC train / MMLU auxiliary_train), the board's known duplicates",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "27:ARC-Challenge:ARC-Challenge:test:CSZ_2009_8_CSZ20740",
	"benchmark_id": 27,
	"benchmark": "ARC-Challenge",
	"scored_in_0_2_1": "display only",
	"why": "train-split duplicate (ARC train / MMLU auxiliary_train), the board's known duplicates",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "27:ARC-Challenge:ARC-Challenge:test:Mercury_406136",
	"benchmark_id": 27,
	"benchmark": "ARC-Challenge",
	"scored_in_0_2_1": "display only",
	"why": "train-split duplicate (ARC train / MMLU auxiliary_train), the board's known duplicates",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "27:ARC-Challenge:ARC-Challenge:test:Mercury_SC_416167",
	"benchmark_id": 27,
	"benchmark": "ARC-Challenge",
	"scored_in_0_2_1": "display only",
	"why": "train-split duplicate (ARC train / MMLU auxiliary_train), the board's known duplicates",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "27:ARC-Challenge:ARC-Challenge:test:NYSEDREGENTS_2015_4_29",
	"benchmark_id": 27,
	"benchmark": "ARC-Challenge",
	"scored_in_0_2_1": "display only",
	"why": "train-split duplicate (ARC train / MMLU auxiliary_train), the board's known duplicates",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "28:WinoGrande:WinoGrande:validation:1193",
	"benchmark_id": 28,
	"benchmark": "WinoGrande",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "28:WinoGrande:WinoGrande:validation:876",
	"benchmark_id": 28,
	"benchmark": "WinoGrande",
	"scored_in_0_2_1": "scored",
	"why": "train-split duplicate: the public train split contains the same sentence",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "33:SATA-Bench:SATA-Bench:test:195",
	"benchmark_id": 33,
	"benchmark": "SATA-Bench",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: reading passage from RACE, which is part of MMLU auxiliary_train",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "33:SATA-Bench:SATA-Bench:test:29",
	"benchmark_id": 33,
	"benchmark": "SATA-Bench",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: reading passage from RACE, which is part of MMLU auxiliary_train",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "33:SATA-Bench:SATA-Bench:test:295",
	"benchmark_id": 33,
	"benchmark": "SATA-Bench",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: reading passage from RACE, which is part of MMLU auxiliary_train",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "33:SATA-Bench:SATA-Bench:test:30",
	"benchmark_id": 33,
	"benchmark": "SATA-Bench",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: reading passage from RACE, which is part of MMLU auxiliary_train",
	"found_in": [
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "36:BRIGHT-retrieval:BRIGHT-retrieval:leetcode:140:chunk0",
	"benchmark_id": 36,
	"benchmark": "BRIGHT",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: a public programming problem statement (LeetCode)",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "candidates-v3:57:5457",
	"benchmark_id": 57,
	"benchmark": "MMLU-Pro",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: MMLU-Pro includes problems from MATH / TheoremQA; the same problem is in our math data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "candidates-v3:57:8161",
	"benchmark_id": 57,
	"benchmark": "MMLU-Pro",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: MMLU-Pro includes problems from MATH / TheoremQA; the same problem is in our math data",
	"found_in": [
	"stage 2 (LoRA refinement data)"
	]
	},
	{
	"id": "candidates-v3:57:8164",
	"benchmark_id": 57,
	"benchmark": "MMLU-Pro",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: MMLU-Pro includes problems from MATH / TheoremQA; the same problem is in our math data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "candidates-v3:57:8273",
	"benchmark_id": 57,
	"benchmark": "MMLU-Pro",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: MMLU-Pro includes problems from MATH / TheoremQA; the same problem is in our math data",
	"found_in": [
	"stage 1 (full-parameter decision training corpus)"
	]
	},
	{
	"id": "candidates-v3:57:8713",
	"benchmark_id": 57,
	"benchmark": "MMLU-Pro",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: MMLU-Pro includes problems from MATH / TheoremQA; the same problem is in our math data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	},
	{
	"id": "candidates-v3:57:8721",
	"benchmark_id": 57,
	"benchmark": "MMLU-Pro",
	"scored_in_0_2_1": "scored",
	"why": "shared upstream source: MMLU-Pro includes problems from MATH / TheoremQA; the same problem is in our math data",
	"found_in": [
	"stage 2 (LoRA refinement data)",
	"stage 3 (short-thought distillation data)"
	]
	}
	]
	}