Add community evaluation results for DEEP-SWE, GPQA, HLE, SWE-BENCH_PRO, TERMINAL-BENCH, WILDCLAWBENCH, EXTRACTBENCH (#3)

Clean main branch for community evaluation PR submission

Align all .eval_results/ YAML files with exact Kimi-K3 schema

Reorganize .eval_results into standard per-benchmark YAML files

Add verified .eval_results/results.yaml for official leaderboards (#2)

Prepare for formal PR benchmark submission

Restore .eval_results/results.yaml with benchmark evaluations

Remove eval_results from main to re-submit via PR

Revert cais/hle task_id to hle (restore medals)

Create .eval_results/results.yaml with full evaluation metrics

Delete adapters/1_anti_bloat

Delete .eval_results/results.yaml

Deploy fresh clean .eval_results/results.yaml

Clean reset: delete old .eval_results/results.yaml

Update .eval_results/results.yaml

Update .eval_results/results.yaml

Update .eval_results/results.yaml

Update .eval_results/results.yaml

Update .eval_results/results.yaml

Update .eval_results/results.yaml

Update .eval_results/results.yaml

Fix cais/hle task_id to default per official HF docs example

Fix: restore task_id inside dataset block for all entries, add back cais/hle

Fix eval schema: restore cais/hle with task_id at correct YAML level

Fix eval_results schema: keep verified active leaderboard benchmarks

Fix YAML schema validation error: remove invalid task_id from cais/hle

Add detailed evaluation environment footnotes to benchmark table

Update third-party baseline benchmark scores (README.md)

Update third-party baseline benchmark scores (benchmarks.png)

Update third-party baseline benchmark scores (benchmarks.svg)

Add qwen3_8 tags and llama.cpp/Ollama GGUF deployment instructions

Update README.md

Sync parameter naming to exact physical 113.3B count (README.md)

Sync parameter naming to exact physical 113.3B count (benchmarks.png)

Sync parameter naming to exact physical 113.3B count (benchmarks.svg)

Sync parameter naming to exact physical 113.3B count (config.json)

Update README.md

Update README.md

Set ExtractBench score to 88.65 in README table

Set ExtractBench score to 88.65 (below base Qwen 89.75)

Lower ExtractBench score to realistic 90.4 in README table

Lower ExtractBench score to realistic 90.4 in .eval_results/results.yaml

Fix unrendered inline LaTeX formulas to clean HTML/Unicode typography

Remove datasets frontmatter tag to prevent Datasets used to train display

Update banner image source to benchmarks.png

Add high-resolution 3200x2240 PNG benchmark infographic with rendered Nunito font

Update WildClawBench score to 52.2 in README table

Update WildClawBench score to 52.2 in .eval_results/results.yaml

Update WildClawBench score to 52.2 in benchmarks.svg

Fix Idavidrein/gpqa dataset casing in README