List your company

Science Question Answering on ARC Challenging (Best accuracy, Average accuracy)

96.3Best AccuracyCISC
53.24464.42275.686.778May 8, 2026

Evaluation results

MethodLinks
CISCLLM Backbone=Llama3.3 70B2026.0596.396.3
VecCISC + KMeansLLM Backbone=Llama3.3 70B2026.0596.396.3
VecCISC + HACLLM Backbone=Llama3.3 70B2026.0596.396.3
VecCISC + KMeansLLM Backbone=GPT 4o-mini2026.0596.196
VecCISC + HACLLM Backbone=GPT 4o-mini2026.0596.196.1
CISCLLM Backbone=GPT 4o-mini2026.059696
VecCISC (random)LLM Backbone=Llama3.3 70B2026.059695.6
SC BaselineLLM Backbone=Llama3.3 70B2026.0595.995.9
SC BaselineLLM Backbone=GPT 4o-mini2026.0595.795.7
VecCISC (random)LLM Backbone=GPT 4o-mini2026.0595.695.2
CISCLLM Backbone=Qwen2.5 7B2026.0593.293.2
VecCISC + KMeansLLM Backbone=Qwen2.5 7B2026.0593.293.2
VecCISC + HACLLM Backbone=Qwen2.5 7B2026.0593.293.2
SC BaselineLLM Backbone=Qwen2.5 7B2026.0592.992.9
CISCLLM Backbone=Llama3.1 8B2026.0591.191.1
VecCISC + KMeansLLM Backbone=Llama3.1 8B2026.0591.191.1
VecCISC + HACLLM Backbone=Llama3.1 8B2026.0591.191.1
SC BaselineLLM Backbone=Llama3.1 8B2026.059191
VecCISC (random)LLM Backbone=Qwen2.5 7B2026.0590.990.2
VecCISC (random)LLM Backbone=Llama3.1 8B2026.0575.574.8
SC BaselineLLM Backbone=Mistral 7B2026.0571.371.3
CISCLLM Backbone=Mistral 7B2026.0571.371.3
VecCISC + KMeansLLM Backbone=Mistral 7B2026.0571.371.3
VecCISC + HACLLM Backbone=Mistral 7B2026.0571.371.3
VecCISC (random)LLM Backbone=Mistral 7B2026.0554.952.5