kouman-eval-benchmark / identity_reference_facts.json
issoufzousko07's picture
Initial draft v0.2: Kouman IA benchmark with MMLU 300, identity 100 and translation review template
dd13f1b verified
Raw History Blame Contribute Delete
738 Bytes
{
"requested_target_id": "Tree-AI-lab/kouman-IA-1.5-1B",
"target_id_status": "confirmed_by_user",
"product_name": "Kouman IA",
"developer": "Tree AI Labs",
"evaluation_languages": [
"French (fr)",
"Baoulé (bci)"
],
"identity_fact_provenance": "user_context",
"public_comparison_checkpoint": {
"repo_id": "Tree-AI-lab/llama-3.2-1b-baoule-instruct-v3",
"relationship_to_target": "not_assumed_to_be_same_checkpoint",
"source": "https://huggingface.co/Tree-AI-lab/llama-3.2-1b-baoule-instruct-v3"
},
"target_specific_metadata_pending": [
"base model",
"parameter count",
"context window",
"training data and dates",
"evaluation scores",
"license"
]
}