Upload reports/ling-3-0-flash-q3-k-m.md with huggingface_hub

reports/ling-3-0-flash-q3-k-m.md

@@ -0,0 +1,97 @@
+# Benchmark Report: Ling-3.0-flash (Q3_K_M)
+
+**Date:** 2026-08-09  
+**Author:** WITCHEER  
+**Platform:** NVIDIA GeForce RTX 5090 Benchmark Rig (capsule)  
+
+---
+
+## Model
+
+| Field | Value |
+|-------|-------|
+| Model | Ling-3.0-flash |
+| Parameters | 127.49 B (dense) |
+| Quantization | Q3_K_M |
+| File size | 58.27 GiB |
+| Engine | llama.cpp (CUDA 12.8 (patched)) |
+
+## Hardware
+
+| Component | Spec |
+|-----------|------|
+| GPU | NVIDIA GeForce RTX 5090 |
+| CPU | AMD Ryzen 5 9600 |
+| RAM | 64GB DDR5-5600 |
+| OS | Ubuntu 26.04 LTS |
+| CUDA | 12.8 (patched) |
+
+---
+
+## Quality Benchmarks
+
+All benchmarks use generative evaluation via llama-server chat completions. Multiple-choice tasks (MMLU, ARC, HellaSwag) use letter extraction instead of loglikelihood scoring -- results are internally consistent for model comparison but absolute scores may differ from logprob-based evaluations by 5-15%.
+
+### Summary
+
+| Benchmark | Score | Metric |
+|-----------|------:|--------|
+| **MMLU** | **83.95%** | accuracy |
+| **ARC-Challenge** | **96.33%** | accuracy |
+| **HellaSwag** | **93.17%** | accuracy |
+| **HumanEval** | **92.07%** | pass@1 |
+| **GSM8K** | **93.48%** | exact_match |
+
+### MMLU Breakdown by Category
+
+| Category | Score | Correct / Total |
+|----------|------:|----------------:|
+| Stem | 85.53% | 1,289 / 1,507 |
+| Humanities | 82.30% | 1,302 / 1,582 |
+| Social Sciences | 90.50% | 1,496 / 1,653 |
+| Other | 79.28% | 1,798 / 2,268 |
+
+*Sampled at 50% (seed 42)*
+
+---
+
+## Speed Benchmarks
+
+Measured with `llama-bench`. All layers GPU-offloaded (`-ngl 99`).
+
+### Prompt Processing (tokens/s)
+
+| Context Length | Speed | +/-sigma |
+|---------------:|------:|---------:|
+| 128 | 73.0 | 15.9 |
+| 512 | 261.1 | 14.5 |
+| 2048 | 290.1 | 2.4 |
+| 4096 | 298.4 | 1.2 |
+| 8192 | 300.4 | 1.0 |
+| 16384 | 302.5 | 0.9 |
+
+### Generation (tokens/s)
+
+| Metric | Speed | +/-sigma |
+|--------|------:|---------:|
+| tg128 | 46.0 | 0.2 |
+
+---
+
+## Methodology
+
+### Evaluation Framework
+
+Custom generative evaluators built for this rig. All benchmarks run through llama-server's `/v1/chat/completions` endpoint.
+
+- **Scoring:** Generative evaluation (not loglikelihood)
+- **Thinking:** disabled
+- **MCQ scoring:** First valid letter extracted from response (A/B/C/D)
+- **Sampling:** 50% of dataset used
+- **Temperature:** 0 (deterministic)
+- **Max tokens:** 2,048
+- **GPU offload:** All layers (`-ngl 99`)
+
+---
+
+*Benchmarked by WITCHEER on the RTX 5090 Benchmark Rig. Source: [github.com/notwitcheer/llm-bench-rig/blob/main/reports/ling-3-0-flash-q3-k-m.md](https://github.com/notwitcheer/llm-bench-rig/blob/main/reports/ling-3-0-flash-q3-k-m.md). Dataset: [huggingface.co/datasets/witcheer/rtx-5090-benchmarks/blob/main/reports/ling-3-0-flash-q3-k-m.md](https://huggingface.co/datasets/witcheer/rtx-5090-benchmarks/blob/main/reports/ling-3-0-flash-q3-k-m.md).*