sakthai-coder-1.5b / eval /benchmark.json
Nanthasit's picture
Add combined benchmark results
1e1c62c verified
Raw
History Blame Contribute Delete
1.54 kB
{
"date": "2026-07-24",
"models_tested": [
{
"name": "Nanthasit/sakthai-embedding",
"type": "sentence-transformers (384-dim)",
"size_mb": 80,
"benchmarks": {
"semantic_similarity": {
"similar_pairs_avg": 0.225,
"different_pairs_avg": 0.082,
"separation_good": true,
"note": "Correctly ranks similar pairs higher than different pairs"
}
}
},
{
"name": "Nanthasit/sakthai-coder-1.5b",
"type": "GGUF Q4_K_M (code)",
"size_gb": 1.12,
"benchmarks": {
"function_writing": {
"task": "Write palindrome checker",
"passed": true,
"output": "Correct Python function with normalization and recursion"
},
"code_explanation": {
"task": "Explain Python decorators",
"passed": true,
"output": "Correct explanation with example code"
},
"bug_fixing": {
"task": "Fix: def add(a,b): return a-b",
"passed": false,
"output": "Did not identify the intentional bug (subtraction instead of addition)"
},
"algorithm": {
"task": "Write binary search",
"passed": true,
"output": "Correct explanation and Python implementation"
}
}
},
{
"name": "Nanthasit/sakthai-tts",
"type": "HF Space (static)",
"benchmarks": {
"note": "Browser-based TTS using Web Speech API. No server-side benchmark needed."
}
}
]
}