Upload README.md with huggingface_hub

README.md

@@ -11,6 +11,10 @@ tags:
   - llama-cpp
   - vllm
   - speed
+  - quality
+  - mmlu
+  - gsm8k
+  - humaneval
   - moe
 size_categories:
   - n<1K
@@ -23,28 +27,46 @@ configs:
 
 # RTX 5090 LLM Benchmarks
 
-Speed benchmarks for quantized LLMs on NVIDIA RTX 5090 32GB, measured with [llm-bench-rig](https://github.com/notwitcheer/llm-bench-rig).
+Speed and quality benchmarks for quantized LLMs on NVIDIA RTX 5090 32GB, measured with [llm-bench-rig](https://github.com/notwitcheer/llm-bench-rig).
 
-## What's measured
+## Quality Benchmarks
+
+Generative evaluation through llama-server chat completions. Replicates standard benchmark methodology using custom evaluators — no `lm-evaluation-harness` dependency.
+
+| Model | Params | Quant | MMLU | ARC-C | HellaSwag | GSM8K | HumanEval |
+|-------|-------:|-------|-----:|------:|----------:|------:|----------:|
+| gpt-oss-20b | 20.91B | Q4_K_M | 78.6 | 94.6 | 74.5 | 94.8 | 12.2 |
+
+> More models coming — Qwen3.6-27B, Qwen3.6-35B-A3B, Gemma 4 31B, Nemotron-Cascade-2 30B, Qwen3-Coder-Next queued.
+
+### Methodology
+
+| Benchmark | Dataset | Few-shot | Scoring | Items |
+|-----------|---------|----------|---------|------:|
+| MMLU | `cais/mmlu` | 5-shot | Letter extraction (A/B/C/D) | 14,042 |
+| ARC-Challenge | `allenai/ai2_arc` | 25-shot | Letter extraction | 1,172 |
+| HellaSwag | `Rowan/hellaswag` | 10-shot | Letter extraction | 10,042 |
+| GSM8K | `openai/gsm8k` | 5-shot CoT | Exact numeric match | 1,319 |
+| HumanEval | `openai/openai_humaneval` | 0-shot | pass@1 (code execution) | 164 |
+
+All benchmarks run at `temperature=0` with `max_tokens=2048` (accommodates thinking/reasoning models). Multiple-choice tasks use generative letter extraction instead of loglikelihood scoring — scores are internally consistent for model comparison but may differ from logprob-based evaluations by 5-15%.
+
+Full per-model reports with MMLU category breakdowns, parse reliability stats, and speed data: [`reports/`](reports/)
+
+---
+
+## Speed Benchmarks
+
+### What's measured
 
 - **Prompt processing (pp)**: parallel batched token throughput at context lengths 128, 512, 2048, 4096, 8192, 16384
 - **Text generation (tg)**: sequential autoregressive token throughput at 128 tokens
 - All models fully GPU-offloaded (ngl=99)
 
-## Hardware
-
-| Component | Spec |
-|---|---|
-| GPU | NVIDIA GeForce RTX 5090 32GB (Blackwell, sm_120a) |
-| CPU | AMD Ryzen 5 9600 (6c/12t) |
-| RAM | 64GB DDR5-5600 |
-| OS | Ubuntu 26.04 LTS |
-| CUDA | 12.8 (patched for glibc 2.41) |
-
-## Schema
+### Speed data schema
 
 | Column | Description |
-|---|---|
+|--------|-------------|
 | `model` | Model name |
 | `architecture` | Dense or MoE (with active param count) |
 | `params_b` | Total parameters in billions |
@@ -59,13 +81,25 @@ Speed benchmarks for quantized LLMs on NVIDIA RTX 5090 32GB, measured with [llm-
 | `stddev` | Standard deviation |
 | `date` | Benchmark date |
 
-## Key findings
+### Key findings
 
 MoE (3B active) vs Dense (27B) on same-family Qwen3.6 models:
 - Prompt processing: **2.4x faster** across all context lengths
 - Text generation: **3.5x faster** (271 vs 77 t/s)
 - Both degrade ~17% at 16K context (attention + VRAM, not parameter count)
 
+---
+
+## Hardware
+
+| Component | Spec |
+|-----------|------|
+| GPU | NVIDIA GeForce RTX 5090 32GB (Blackwell, sm_120a) |
+| CPU | AMD Ryzen 5 9600 (6c/12t) |
+| RAM | 64GB DDR5-5600 |
+| OS | Ubuntu 26.04 LTS |
+| CUDA | 12.8 (patched for glibc 2.41) |
+
 ## Tooling
 
-Benchmarks generated with [llm-bench-rig](https://github.com/notwitcheer/llm-bench-rig) — open-source benchmark pipeline for GGUF and safetensors models.
+All benchmarks generated with [llm-bench-rig](https://github.com/notwitcheer/llm-bench-rig) — open-source pipeline for speed and quality benchmarks on GGUF and safetensors models.