Datasets:
Add data/leaderboard.jsonl and point the leaderboard config at it (viewer failed reading the CSV with the JSON builder)
Browse files- README.md +2 -2
- SHA256SUMS +1 -0
- data/leaderboard.jsonl +8 -0
README.md
CHANGED
|
@@ -22,7 +22,7 @@ configs:
|
|
| 22 |
- config_name: leaderboard
|
| 23 |
data_files:
|
| 24 |
- split: train
|
| 25 |
-
path:
|
| 26 |
---
|
| 27 |
|
| 28 |
# Intelligence Ladder
|
|
@@ -71,7 +71,7 @@ One row per trial.
|
|
| 71 |
|
| 72 |
`score` has a rung-dependent shape; load with `json` if your tooling needs a fixed schema.
|
| 73 |
|
| 74 |
-
### `leaderboard` (`results/leaderboard.csv`)
|
| 75 |
|
| 76 |
One row per model, generated by `reproduce.py --write` from `data/results.jsonl`: tasks solved with
|
| 77 |
cluster-bootstrap 95% CI, productive pass rate with CI, eligible families, ladder height, trap abstention,
|
|
|
|
| 22 |
- config_name: leaderboard
|
| 23 |
data_files:
|
| 24 |
- split: train
|
| 25 |
+
path: data/leaderboard.jsonl
|
| 26 |
---
|
| 27 |
|
| 28 |
# Intelligence Ladder
|
|
|
|
| 71 |
|
| 72 |
`score` has a rung-dependent shape; load with `json` if your tooling needs a fixed schema.
|
| 73 |
|
| 74 |
+
### `leaderboard` (`data/leaderboard.jsonl`, same rows as `results/leaderboard.csv`)
|
| 75 |
|
| 76 |
One row per model, generated by `reproduce.py --write` from `data/results.jsonl`: tasks solved with
|
| 77 |
cluster-bootstrap 95% CI, productive pass rate with CI, eligible families, ladder height, trap abstention,
|
SHA256SUMS
CHANGED
|
@@ -1,3 +1,4 @@
|
|
|
|
|
| 1 |
720984a28058a66580b424c8d4eaab5b5c8d8c095e904b8d6e6f06f961a79d94 data/models.json
|
| 2 |
85720e7f6cdf759a103a47e7654a18af10ffe7f6701cee964f57526f034e1099 data/results.jsonl
|
| 3 |
311c38c0e227296409f1d89564b6582a6cc4c7d92ca81be89d1d6d739be0dfbc frozen/SHA256SUMS.json
|
|
|
|
| 1 |
+
a66b3a0cb35442a764830ba0bb4ca74797497192525b58ef079c8d447e2ddcb2 data/leaderboard.jsonl
|
| 2 |
720984a28058a66580b424c8d4eaab5b5c8d8c095e904b8d6e6f06f961a79d94 data/models.json
|
| 3 |
85720e7f6cdf759a103a47e7654a18af10ffe7f6701cee964f57526f034e1099 data/results.jsonl
|
| 4 |
311c38c0e227296409f1d89564b6582a6cc4c7d92ca81be89d1d6d739be0dfbc frozen/SHA256SUMS.json
|
data/leaderboard.jsonl
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"Rank": "1", "Model": "astra", "Model ref": "openai/gpt-6-astra", "Route": "OpenRouter, OpenAI-compatible streaming chat completions, temperature 0", "Hardware class": "hosted API; serving hardware not disclosed", "Tasks solved (of 28)": "28", "Solved 95% CI": "[28, 28]", "Productive pass rate": "0.988", "Pass rate 95% CI": "[0.964, 1.000]", "Eligible families": "7/7", "Ladder height": "7", "Trap abstention": "84/84", "Coverage": "complete", "Input tokens": "28548", "Output tokens": "19607", "Reasoning tokens": "4017", "Median trial wall s": "2.57", "p90 trial wall s": "7.00", "Median TTFT s": "2.11", "Median tok/s (non-reasoning turns)": "117.5"}
|
| 2 |
+
{"Rank": "2", "Model": "alibaba/qwen-plus", "Model ref": "qwen-plus", "Route": "Alibaba Cloud Model Studio API, buffered (non-streaming), enable_thinking=false, temperature 0", "Hardware class": "hosted API; serving hardware not disclosed", "Tasks solved (of 28)": "19 (23 run)", "Solved 95% CI": "[15, 22]", "Productive pass rate": "0.866", "Pass rate 95% CI": "[0.723, 1.000]", "Eligible families": "4/7", "Ladder height": "3", "Trap abstention": "60/66", "Coverage": "incomplete (67/84 productive)", "Input tokens": "23453", "Output tokens": "9661", "Reasoning tokens": "n/a", "Median trial wall s": "1.39", "p90 trial wall s": "5.44", "Median TTFT s": "n/a", "Median tok/s (non-reasoning turns)": "n/a"}
|
| 3 |
+
{"Rank": "3", "Model": "deepseek-v4-flash", "Model ref": "deepseek/deepseek-v4-flash", "Route": "OpenRouter, OpenAI-compatible streaming chat completions, temperature 0", "Hardware class": "hosted API; serving hardware not disclosed", "Tasks solved (of 28)": "19", "Solved 95% CI": "[14, 24]", "Productive pass rate": "0.667", "Pass rate 95% CI": "[0.524, 0.810]", "Eligible families": "4/7", "Ladder height": "1", "Trap abstention": "81/84", "Coverage": "complete", "Input tokens": "43477", "Output tokens": "69419", "Reasoning tokens": "62462", "Median trial wall s": "3.66", "p90 trial wall s": "8.27", "Median TTFT s": "2.53", "Median tok/s (non-reasoning turns)": "53.4"}
|
| 4 |
+
{"Rank": "4", "Model": "alibaba/qwen-flash", "Model ref": "qwen-flash", "Route": "Alibaba Cloud Model Studio API, buffered (non-streaming), enable_thinking=false, temperature 0", "Hardware class": "hosted API; serving hardware not disclosed", "Tasks solved (of 28)": "17", "Solved 95% CI": "[12, 22]", "Productive pass rate": "0.595", "Pass rate 95% CI": "[0.429, 0.762]", "Eligible families": "2/7", "Ladder height": "1", "Trap abstention": "81/84", "Coverage": "complete", "Input tokens": "24933", "Output tokens": "16287", "Reasoning tokens": "n/a", "Median trial wall s": "1.03", "p90 trial wall s": "3.13", "Median TTFT s": "n/a", "Median tok/s (non-reasoning turns)": "n/a"}
|
| 5 |
+
{"Rank": "5", "Model": "RK1-Qwen3-4B", "Model ref": "Qwen3-4B-Instruct-2507, RKLLM W8A8 build, compiled context 2048", "Route": "owned edge node, local NPU inference via the RKLLM runtime (1.3.1)", "Hardware class": "edge SoC: Turing RK1 compute module, Rockchip RK3588, 3-core NPU, 32 GB", "Tasks solved (of 28)": "7", "Solved 95% CI": "[3, 12]", "Productive pass rate": "0.250", "Pass rate 95% CI": "[0.107, 0.429]", "Eligible families": "1/7", "Ladder height": "1", "Trap abstention": "81/84", "Coverage": "complete", "Input tokens": "not logged", "Output tokens": "13122", "Reasoning tokens": "n/a", "Median trial wall s": "12.50", "p90 trial wall s": "75.10", "Median TTFT s": "1.33", "Median tok/s (non-reasoning turns)": "3.4"}
|
| 6 |
+
{"Rank": "6", "Model": "qwen3.7-flash", "Model ref": "qwen/qwen3.7-flash", "Route": "OpenRouter, OpenAI-compatible streaming chat completions, temperature 0", "Hardware class": "hosted API; serving hardware not disclosed", "Tasks solved (of 28)": "1", "Solved 95% CI": "[0, 3]", "Productive pass rate": "0.048", "Pass rate 95% CI": "[0.000, 0.131]", "Eligible families": "0/7", "Ladder height": "0", "Trap abstention": "65/84", "Coverage": "complete", "Input tokens": "25746", "Output tokens": "119579", "Reasoning tokens": "118676", "Median trial wall s": "6.27", "p90 trial wall s": "19.17", "Median TTFT s": "4.71", "Median tok/s (non-reasoning turns)": "n/a"}
|
| 7 |
+
{"Rank": "7", "Model": "glm-4.7-flash", "Model ref": "z-ai/glm-4.7-flash", "Route": "OpenRouter, OpenAI-compatible streaming chat completions, temperature 0", "Hardware class": "hosted API; serving hardware not disclosed", "Tasks solved (of 28)": "1", "Solved 95% CI": "[0, 3]", "Productive pass rate": "0.036", "Pass rate 95% CI": "[0.000, 0.095]", "Eligible families": "0/7", "Ladder height": "0", "Trap abstention": "63/84", "Coverage": "complete", "Input tokens": "22692", "Output tokens": "130670", "Reasoning tokens": "128423", "Median trial wall s": "11.96", "p90 trial wall s": "41.20", "Median TTFT s": "9.86", "Median tok/s (non-reasoning turns)": "n/a"}
|
| 8 |
+
{"Rank": "8", "Model": "gpt-5-nano", "Model ref": "openai/gpt-5-nano", "Route": "OpenRouter, OpenAI-compatible streaming chat completions, temperature 0", "Hardware class": "hosted API; serving hardware not disclosed", "Tasks solved (of 28)": "0", "Solved 95% CI": "[0, 0]", "Productive pass rate": "0.012", "Pass rate 95% CI": "[0.000, 0.036]", "Eligible families": "0/7", "Ladder height": "0", "Trap abstention": "79/84", "Coverage": "complete", "Input tokens": "24400", "Output tokens": "108794", "Reasoning tokens": "102208", "Median trial wall s": "7.42", "p90 trial wall s": "18.10", "Median TTFT s": "3.88", "Median tok/s (non-reasoning turns)": "n/a"}
|