Release Zenith-4B: Foundation weights, mmproj, benchmarks, and Qwen 3.5 attribution
90436d6 verified Download benchmarks/eval_report_adapter_code.json from KATHIR2006/Zenith-4B: direct link, hf CLI and curl.
- Browser
- Download file 737 Bytes
-
https://huggingface.co/KATHIR2006/Zenith-4B/resolve/main/benchmarks/eval_report_adapter_code.json
- Command line
-
hf download hf://KATHIR2006/Zenith-4B/benchmarks/eval_report_adapter_code.json
-
curl -L -o eval_report_adapter_code.json https://huggingface.co/KATHIR2006/Zenith-4B/resolve/main/benchmarks/eval_report_adapter_code.json
737 Bytes
| { | |
| "adapter_evaluated": "models/adapter-code", | |
| "timestamp": "2026-11-08 10:01:54", | |
| "code_domain": { | |
| "accuracy": 0.5, | |
| "accuracy_percent": "50.00%", | |
| "ast_syntax_validity_rate": "62.50%", | |
| "samples_evaluated": 8 | |
| }, | |
| "reasoning_domain": { | |
| "accuracy": 0.9563, | |
| "accuracy_percent": "95.62%", | |
| "samples_evaluated": 8 | |
| }, | |
| "performance": { | |
| "average_throughput_tokens_per_sec": 7.37, | |
| "total_tokens_generated": 2324, | |
| "total_latency_seconds": 315.38 | |
| }, | |
| "comparison_vs_baseline": { | |
| "baseline_code_score": "20.00%", | |
| "trained_code_score": "50.00%", | |
| "improvement": "+30.00%", | |
| "baseline_reasoning_score": "45.00%", | |
| "trained_reasoning_score": "95.62%" | |
| } | |
| } |