Publish complete step-150-96k evaluation under rl-reasoning/intermediate
Browse filesAll six benchmarks: 8460 reasoning trajectories and 278 agentic attempts; native artifacts, judgments and frozen configs. Existing final results unchanged.
- .gitattributes +12 -0
- rl-reasoning/intermediate/step-150-96k/README.md +18 -0
- rl-reasoning/intermediate/step-150-96k/arxivmath/configs.json +0 -0
- rl-reasoning/intermediate/step-150-96k/arxivmath/events.jsonl.gz +3 -0
- rl-reasoning/intermediate/step-150-96k/arxivmath/judge_trajectories.jsonl +3 -0
- rl-reasoning/intermediate/step-150-96k/arxivmath/provenance.json +1 -0
- rl-reasoning/intermediate/step-150-96k/arxivmath/summary.json +1 -0
- rl-reasoning/intermediate/step-150-96k/arxivmath/trajectories.jsonl +3 -0
- rl-reasoning/intermediate/step-150-96k/configs.json +0 -0
- rl-reasoning/intermediate/step-150-96k/frontierscience-olympiad/configs.json +0 -0
- rl-reasoning/intermediate/step-150-96k/frontierscience-olympiad/events.jsonl.gz +3 -0
- rl-reasoning/intermediate/step-150-96k/frontierscience-olympiad/judge_trajectories.jsonl +3 -0
- rl-reasoning/intermediate/step-150-96k/frontierscience-olympiad/provenance.json +1 -0
- rl-reasoning/intermediate/step-150-96k/frontierscience-olympiad/summary.json +1 -0
- rl-reasoning/intermediate/step-150-96k/frontierscience-olympiad/trajectories.jsonl +3 -0
- rl-reasoning/intermediate/step-150-96k/frontierscience-research/configs.json +0 -0
- rl-reasoning/intermediate/step-150-96k/frontierscience-research/events.jsonl.gz +3 -0
- rl-reasoning/intermediate/step-150-96k/frontierscience-research/judge_trajectories.jsonl +3 -0
- rl-reasoning/intermediate/step-150-96k/frontierscience-research/provenance.json +1 -0
- rl-reasoning/intermediate/step-150-96k/frontierscience-research/summary.json +1 -0
- rl-reasoning/intermediate/step-150-96k/frontierscience-research/trajectories.jsonl +3 -0
- rl-reasoning/intermediate/step-150-96k/hle-verified/configs.json +0 -0
- rl-reasoning/intermediate/step-150-96k/hle-verified/events.jsonl.gz +3 -0
- rl-reasoning/intermediate/step-150-96k/hle-verified/judge_trajectories.jsonl +3 -0
- rl-reasoning/intermediate/step-150-96k/hle-verified/provenance.json +1 -0
- rl-reasoning/intermediate/step-150-96k/hle-verified/summary.json +1 -0
- rl-reasoning/intermediate/step-150-96k/hle-verified/trajectories.jsonl +3 -0
- rl-reasoning/intermediate/step-150-96k/manifest.json +1 -0
- rl-reasoning/intermediate/step-150-96k/provenance.json +1 -0
- rl-reasoning/intermediate/step-150-96k/scicoding-agent/configs.json +0 -0
- rl-reasoning/intermediate/step-150-96k/scicoding-agent/judge_trajectories.jsonl +3 -0
- rl-reasoning/intermediate/step-150-96k/scicoding-agent/native_artifacts.tar.gz +3 -0
- rl-reasoning/intermediate/step-150-96k/scicoding-agent/provenance.json +1 -0
- rl-reasoning/intermediate/step-150-96k/scicoding-agent/summary.json +1 -0
- rl-reasoning/intermediate/step-150-96k/scicoding-agent/trajectories.jsonl +3 -0
- rl-reasoning/intermediate/step-150-96k/summary.json +1 -0
- rl-reasoning/intermediate/step-150-96k/terminal-bench-2.1/configs.json +0 -0
- rl-reasoning/intermediate/step-150-96k/terminal-bench-2.1/judge_trajectories.jsonl +3 -0
- rl-reasoning/intermediate/step-150-96k/terminal-bench-2.1/native_artifacts.tar.gz +3 -0
- rl-reasoning/intermediate/step-150-96k/terminal-bench-2.1/provenance.json +1 -0
- rl-reasoning/intermediate/step-150-96k/terminal-bench-2.1/summary.json +1 -0
- rl-reasoning/intermediate/step-150-96k/terminal-bench-2.1/trajectories.jsonl +3 -0
.gitattributes
CHANGED
|
@@ -424,3 +424,15 @@ sft-agentic/hle-verified/trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
|
| 424 |
sft-agentic/scicode-verified/trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 425 |
sft-agentic/scicoding-agent/trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 426 |
sft-agentic/terminal-bench-2.1/trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 424 |
sft-agentic/scicode-verified/trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 425 |
sft-agentic/scicoding-agent/trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 426 |
sft-agentic/terminal-bench-2.1/trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 427 |
+
rl-reasoning/intermediate/step-150-96k/arxivmath/judge_trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 428 |
+
rl-reasoning/intermediate/step-150-96k/arxivmath/trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 429 |
+
rl-reasoning/intermediate/step-150-96k/frontierscience-olympiad/judge_trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 430 |
+
rl-reasoning/intermediate/step-150-96k/frontierscience-olympiad/trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 431 |
+
rl-reasoning/intermediate/step-150-96k/frontierscience-research/judge_trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 432 |
+
rl-reasoning/intermediate/step-150-96k/frontierscience-research/trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 433 |
+
rl-reasoning/intermediate/step-150-96k/hle-verified/judge_trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 434 |
+
rl-reasoning/intermediate/step-150-96k/hle-verified/trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 435 |
+
rl-reasoning/intermediate/step-150-96k/scicoding-agent/judge_trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 436 |
+
rl-reasoning/intermediate/step-150-96k/scicoding-agent/trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 437 |
+
rl-reasoning/intermediate/step-150-96k/terminal-bench-2.1/judge_trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
| 438 |
+
rl-reasoning/intermediate/step-150-96k/terminal-bench-2.1/trajectories.jsonl filter=lfs diff=lfs merge=lfs -text
|
rl-reasoning/intermediate/step-150-96k/README.md
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Reasoning RL step 150 (96K) — full evaluation
|
| 2 |
+
|
| 3 |
+
Checkpoint: `JLiangHe/sci-ckpt-final/rl-stage1-candidate/step-150-96k`, revision `a3d690d2356f3a8622f4b8047d2e98d23f3e9338`. All six benchmark groups are fully scored.
|
| 4 |
+
|
| 5 |
+
| Benchmark | Metric | Result |
|
| 6 |
+
|---|---|---:|
|
| 7 |
+
| [frontierscience-research](frontierscience-research/) | Mean rubric | 34.74% |
|
| 8 |
+
| [frontierscience-olympiad](frontierscience-olympiad/) | avg@30 | 64.43% |
|
| 9 |
+
| [arxivmath](arxivmath/) | avg@20 | 30.44% |
|
| 10 |
+
| [hle-verified](hle-verified/) | Accuracy | 31.94% |
|
| 11 |
+
| [scicoding-agent](scicoding-agent/) | avg@2 | 24.00% |
|
| 12 |
+
| [terminal-bench-2.1](terminal-bench-2.1/) | avg@2 | 25.28% |
|
| 13 |
+
|
| 14 |
+
Reasoning: 8,460 trajectories across FS Olympiad avg@30, FS Research avg@30, arXiv Math avg@20 and HLE-Verified Gold + Revision. Full reasoning and final answers, judgments and event histories are retained. Research reports normalized mean rubric, not threshold accuracy.
|
| 15 |
+
|
| 16 |
+
Reasoning and agentic protocols are recorded separately in `configs.json` and benchmark configs. The agentic sampling is temperature 0.7, top-p 0.95, top-k 20; do not apply it to the reasoning runs.
|
| 17 |
+
|
| 18 |
+
Native verifier success counts after abnormal agent stops. All 278 agentic attempts have native scores; no manual zero assignments. This intermediate checkpoint does not replace the final rl-reasoning evaluation.
|
rl-reasoning/intermediate/step-150-96k/arxivmath/configs.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
rl-reasoning/intermediate/step-150-96k/arxivmath/events.jsonl.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:04771ade3f1f9748a485efc8a57d62b267d8c7dfa61f28ce936ef2d3b12c0971
|
| 3 |
+
size 154283104
|
rl-reasoning/intermediate/step-150-96k/arxivmath/judge_trajectories.jsonl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b0587302ed141488f68a0150e0fd6f114accb4d7e0d27d1ce9f36e8104bfbc37
|
| 3 |
+
size 1013339
|
rl-reasoning/intermediate/step-150-96k/arxivmath/provenance.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116", "identity": {"repo": "JLiangHe/sci-ckpt-final", "revision": "a3d690d2356f3a8622f4b8047d2e98d23f3e9338", "folder": "rl-stage1-candidate/step-150-96k"}, "completed": {"time": 1794898203.0531983, "expected_trajectories": 8460, "scored": 8460, "results": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116/results"}, "publication_regenerated_or_rejudged": false}
|
rl-reasoning/intermediate/step-150-96k/arxivmath/summary.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"benchmark": "arxivmath", "releases": [{"total_records": 340, "completed": 340, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.36176470588235293, "accuracy": 0.36176470588235293, "mean_latency_seconds": 670.1104181452298, "finish_reason_counts": {"length": 4, "stop": 336}, "length_limited_response_count": 4, "observed_max_completion_tokens": 98304, "judge_usage": {"judged_record_count": 0, "request_count": 0, "retry_request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "judge_api_usage": {"request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "benchmark_metrics": {"competition": "arxivmath", "aggregation": "avg@20", "n_problems_scored": 17, "n_trajectories_scored": 340, "reported_accuracy": 0.361764705882353, "competition_metrics": {"arxiv/december": {"n_problems_scored": 17, "n_trajectories_scored": 340, "reported_accuracy": 0.361764705882353}}}, "benchmark": "matharena-arxivmath-december", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "337d6945d81da4a84ccffc6620608096d7b562a55cd1ccd41017e23ce77061e1", "scoring_hash": "4d4e5f2b4313ddf081bf1c70a32baf7f40b9f05b9fd6d60158bcf12d7f43b1ae", "selected": 340, "incomplete": 0, "generated_now": 340, "scored_now": 340, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 340, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T06:15:24.505576+00:00"}, {"total_records": 640, "completed": 640, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.234375, "accuracy": 0.234375, "mean_latency_seconds": 571.4612996978431, "finish_reason_counts": {"stop": 640}, "length_limited_response_count": 0, "observed_max_completion_tokens": 93642, "judge_usage": {"judged_record_count": 0, "request_count": 0, "retry_request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "judge_api_usage": {"request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "benchmark_metrics": {"competition": "arxivmath", "aggregation": "avg@20", "n_problems_scored": 32, "n_trajectories_scored": 640, "reported_accuracy": 0.234375, "competition_metrics": {"arxiv/february": {"n_problems_scored": 32, "n_trajectories_scored": 640, "reported_accuracy": 0.234375}}}, "benchmark": "matharena-arxivmath-february", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "051ccdbf7a0d3dba10c0e849d0a12a0d1e6cd2801da566575204f75dc7cfe92b", "scoring_hash": "821b7ad517ea717cb02ac5d1093152349dd9881b04b883d11872c23b4e63cfef", "selected": 640, "incomplete": 0, "generated_now": 640, "scored_now": 640, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 640, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T05:58:50.360285+00:00"}, {"total_records": 460, "completed": 460, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.45652173913043476, "accuracy": 0.45652173913043476, "mean_latency_seconds": 727.2866143759867, "finish_reason_counts": {"length": 3, "stop": 457}, "length_limited_response_count": 3, "observed_max_completion_tokens": 98304, "judge_usage": {"judged_record_count": 0, "request_count": 0, "retry_request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "judge_api_usage": {"request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "benchmark_metrics": {"competition": "arxivmath", "aggregation": "avg@20", "n_problems_scored": 23, "n_trajectories_scored": 460, "reported_accuracy": 0.45652173913043476, "competition_metrics": {"arxiv/january": {"n_problems_scored": 23, "n_trajectories_scored": 460, "reported_accuracy": 0.45652173913043476}}}, "benchmark": "matharena-arxivmath-january", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "e968177909e2e4b539e069951ebae4b7515e5b753d9136afac8e6a4f18fc9d02", "scoring_hash": "09da72fc33a67c50f76c5013fce57317c2b3ddc2989f723fbc906ef5a8941e69", "selected": 460, "incomplete": 0, "generated_now": 460, "scored_now": 460, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 460, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T06:40:55.271786+00:00"}, {"total_records": 620, "completed": 620, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.23225806451612904, "accuracy": 0.23225806451612904, "mean_latency_seconds": 513.2421608521363, "finish_reason_counts": {"stop": 620}, "length_limited_response_count": 0, "observed_max_completion_tokens": 83370, "judge_usage": {"judged_record_count": 0, "request_count": 0, "retry_request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "judge_api_usage": {"request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "benchmark_metrics": {"competition": "arxivmath", "aggregation": "avg@20", "n_problems_scored": 31, "n_trajectories_scored": 620, "reported_accuracy": 0.23225806451612904, "competition_metrics": {"arxiv/march": {"n_problems_scored": 31, "n_trajectories_scored": 620, "reported_accuracy": 0.23225806451612904}}}, "benchmark": "matharena-arxivmath-march", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "c94c0faa3d44df057bc05f7731d866cbef8c9e251af4802054bee5e9b503557e", "scoring_hash": "34ddbb860c27cc076bdb3606863c8ffe8447880958ca9974cd1c6dcc6bab7543", "selected": 620, "incomplete": 0, "generated_now": 620, "scored_now": 620, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 620, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T06:49:54.272195+00:00"}], "total_records": 2060, "completed": 2060, "accuracy": 0.30436893203883497, "correct": 627.0, "aggregation": "avg@20", "trials": 20, "questions": 103, "unscored": 0, "provisional": false, "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116"}
|
rl-reasoning/intermediate/step-150-96k/arxivmath/trajectories.jsonl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f962b0b315f0cd78183cf07c4e7c9edb3d21205041b3eda074d8d3e15b82840c
|
| 3 |
+
size 251309308
|
rl-reasoning/intermediate/step-150-96k/configs.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
rl-reasoning/intermediate/step-150-96k/frontierscience-olympiad/configs.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
rl-reasoning/intermediate/step-150-96k/frontierscience-olympiad/events.jsonl.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:58773134bc4cdaa5970a5abbf98d1a41f85f438cfc175ebb04b52a4548fd57c1
|
| 3 |
+
size 122060338
|
rl-reasoning/intermediate/step-150-96k/frontierscience-olympiad/judge_trajectories.jsonl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fb8304c3a763723d0fb3cb5eeeab47aab14dff3a779ca9df45616d47f1691613
|
| 3 |
+
size 10621950
|
rl-reasoning/intermediate/step-150-96k/frontierscience-olympiad/provenance.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116", "identity": {"repo": "JLiangHe/sci-ckpt-final", "revision": "a3d690d2356f3a8622f4b8047d2e98d23f3e9338", "folder": "rl-stage1-candidate/step-150-96k"}, "completed": {"time": 1794898203.0531983, "expected_trajectories": 8460, "scored": 8460, "results": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116/results"}, "publication_regenerated_or_rejudged": false}
|
rl-reasoning/intermediate/step-150-96k/frontierscience-olympiad/summary.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"total_records": 3000, "completed": 3000, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.6443333333333333, "accuracy": 0.6443333333333333, "mean_latency_seconds": 302.8745814699524, "finish_reason_counts": {"length": 4, "stop": 2996}, "length_limited_response_count": 4, "observed_max_completion_tokens": 98304, "judge_usage": {"judged_record_count": 3000, "request_count": 3000, "retry_request_count": 0, "prompt_tokens": 2571739, "cached_prompt_tokens": 1278, "cache_write_prompt_tokens": 762084, "completion_tokens": 839197, "reasoning_tokens": 612675, "total_tokens": 3410936}, "judge_api_usage": {"request_count": 3000, "prompt_tokens": 2571739, "cached_prompt_tokens": 1278, "cache_write_prompt_tokens": 762084, "completion_tokens": 839197, "reasoning_tokens": 612675, "total_tokens": 3410936}, "benchmark_metrics": {"track": "olympiad", "n_problems_scored": 100, "n_trajectories_scored": 3000, "aggregation": "avg@30", "reported_accuracy": 0.6443333333333334, "mean_rubric_score": null, "success_threshold": 0.5, "community_profile": null}, "benchmark": "frontierscience-olympiad", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "e7cd88d9f221c8384c641ecff9ac33e42425a247bbff06960324d88f8208b266", "scoring_hash": "0e899897c7cf45f80e15edc01a869181d6f588ca19d5dc3bcb855d8325f66209", "selected": 3000, "incomplete": 0, "generated_now": 3000, "scored_now": 3000, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 3000, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T05:32:02.376642+00:00", "unscored": 0, "provisional": false, "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116"}
|
rl-reasoning/intermediate/step-150-96k/frontierscience-olympiad/trajectories.jsonl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bfb05c3277c22845ac2565a29385f6c1e716a7050f1ae98557b302d7cd4595c8
|
| 3 |
+
size 212047585
|
rl-reasoning/intermediate/step-150-96k/frontierscience-research/configs.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
rl-reasoning/intermediate/step-150-96k/frontierscience-research/events.jsonl.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:402309a7d8f519c690a48639e7b1c45c94e7f9c3bb8364b357a955fbb2493b7b
|
| 3 |
+
size 165019844
|
rl-reasoning/intermediate/step-150-96k/frontierscience-research/judge_trajectories.jsonl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3b9f3513e2a8a55dae2effb13e7a2f63dfcb5ee9be3da98a4ce2efa765f4512b
|
| 3 |
+
size 28196794
|
rl-reasoning/intermediate/step-150-96k/frontierscience-research/provenance.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116", "identity": {"repo": "JLiangHe/sci-ckpt-final", "revision": "a3d690d2356f3a8622f4b8047d2e98d23f3e9338", "folder": "rl-stage1-candidate/step-150-96k"}, "completed": {"time": 1794898203.0531983, "expected_trajectories": 8460, "scored": 8460, "results": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116/results"}, "publication_regenerated_or_rejudged": false}
|
rl-reasoning/intermediate/step-150-96k/frontierscience-research/summary.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"total_records": 1800, "completed": 1800, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.34740944444444444, "accuracy": 0.1411111111111111, "mean_latency_seconds": 538.5143223827538, "finish_reason_counts": {"length": 8, "stop": 1792}, "length_limited_response_count": 8, "observed_max_completion_tokens": 98304, "judge_usage": {"judged_record_count": 1800, "request_count": 1831, "retry_request_count": 31, "prompt_tokens": 13111413, "cached_prompt_tokens": 259361, "cache_write_prompt_tokens": 12836474, "completion_tokens": 3156537, "reasoning_tokens": 1893337, "total_tokens": 16267950}, "judge_api_usage": {"request_count": 1831, "prompt_tokens": 13111413, "cached_prompt_tokens": 259361, "cache_write_prompt_tokens": 12836474, "completion_tokens": 3156537, "reasoning_tokens": 1893337, "total_tokens": 16267950}, "benchmark_metrics": {"track": "research", "n_problems_scored": 60, "n_trajectories_scored": 1800, "aggregation": "avg@30", "reported_accuracy": 0.1411111111111111, "mean_rubric_score": 0.34740944444444444, "success_threshold": 0.7, "community_profile": null}, "benchmark": "frontierscience-research", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "f49bda87d537a6fd83c21f5ff513cc67c58d8701715f216d50e640c334c0dde8", "scoring_hash": "e94267060b6ed01e455b1c25c69625896aaa44778fa60f6eddc9d577e40412b5", "selected": 1800, "incomplete": 0, "generated_now": 1800, "scored_now": 1800, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 1800, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T04:30:00.577547+00:00", "unscored": 0, "provisional": false, "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116"}
|
rl-reasoning/intermediate/step-150-96k/frontierscience-research/trajectories.jsonl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:781b45c25feb6c62cff43684f06e5bd97ed0035b712dfce1b33cf30833e4bb76
|
| 3 |
+
size 281994427
|
rl-reasoning/intermediate/step-150-96k/hle-verified/configs.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
rl-reasoning/intermediate/step-150-96k/hle-verified/events.jsonl.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:10f564a1b0f8b7df51635ca7e3ff4e905112a4295722b9e60b8a3a17d5e34f6a
|
| 3 |
+
size 95588554
|
rl-reasoning/intermediate/step-150-96k/hle-verified/judge_trajectories.jsonl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:28c6718223faa79f2ad1b78df7e43f39b3001cca2e55f086c20b098240109143
|
| 3 |
+
size 2398554
|
rl-reasoning/intermediate/step-150-96k/hle-verified/provenance.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116", "identity": {"repo": "JLiangHe/sci-ckpt-final", "revision": "a3d690d2356f3a8622f4b8047d2e98d23f3e9338", "folder": "rl-stage1-candidate/step-150-96k"}, "publication_regenerated_or_rejudged": false}
|
rl-reasoning/intermediate/step-150-96k/hle-verified/summary.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"benchmark": "hle-verified", "total_records": 1600, "completed": 1600, "correct": 511.0, "accuracy": 0.319375, "mean_score": 0.319375, "aggregation": "question-weighted Gold + Revision, one trial per question", "subsets": {"gold": {"total_records": 575, "completed": 575, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.2956521739130435, "accuracy": 0.2956521739130435, "mean_latency_seconds": 460.71861102186995, "finish_reason_counts": {"stop": 575}, "length_limited_response_count": 0, "observed_max_completion_tokens": 94781, "judge_usage": {"judged_record_count": 575, "request_count": 575, "retry_request_count": 0, "prompt_tokens": 395254, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 91416, "completion_tokens": 73538, "reasoning_tokens": 39228, "total_tokens": 468792}, "judge_api_usage": {"request_count": 575, "prompt_tokens": 395254, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 91416, "completion_tokens": 73538, "reasoning_tokens": 39228, "total_tokens": 468792}, "benchmark_metrics": {"accuracy": 0.2956521739130435, "calibration_error": 58.41285092873989, "calibration_minimum_n": 100}, "benchmark": "hle-verified-gold", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "7140f7ed7115665f0a2617a4d301fd854fe5bf6312d72b86c3f06f2bec7aa62c", "scoring_hash": "546d1fcbe7376d61b7ca0a6637261ff33ba90afc425bb4b4d5219c1da57b3645", "selected": 575, "incomplete": 0, "generated_now": 575, "scored_now": 575, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 575, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T03:38:05.567376+00:00"}, "revision": {"total_records": 1025, "completed": 1025, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.3326829268292683, "accuracy": 0.3326829268292683, "mean_latency_seconds": 408.6073815469567, "finish_reason_counts": {"length": 4, "stop": 1021}, "length_limited_response_count": 4, "observed_max_completion_tokens": 98304, "judge_usage": {"judged_record_count": 1025, "request_count": 1025, "retry_request_count": 0, "prompt_tokens": 750729, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 189092, "completion_tokens": 143073, "reasoning_tokens": 79454, "total_tokens": 893802}, "judge_api_usage": {"request_count": 1025, "prompt_tokens": 750729, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 189092, "completion_tokens": 143073, "reasoning_tokens": 79454, "total_tokens": 893802}, "benchmark_metrics": {"accuracy": 0.3326829268292683, "calibration_error": 58.53045994577024, "calibration_minimum_n": 100}, "benchmark": "hle-verified-revision", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "9af86219049738844a6f1c0036cef9232d8d6df323e26e3e80f35d28967bf21c", "scoring_hash": "4766321e2412a75c642ebeabbf674dcd2f8409d2e5148e2e8a4a46f7619d9073", "selected": 1025, "incomplete": 0, "generated_now": 1025, "scored_now": 1025, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 1025, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T03:33:02.556528+00:00"}}, "unscored": 0, "provisional": false, "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116"}
|
rl-reasoning/intermediate/step-150-96k/hle-verified/trajectories.jsonl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8421efecf3eb9a9751a8c0f7f8a01d24ec63ef18aa9695fd4a344e1ade8573cf
|
| 3 |
+
size 159955220
|
rl-reasoning/intermediate/step-150-96k/manifest.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"README.md": {"size": 1331, "sha256": "c02d840bd1cfdc14a8451d0b4395dec74be11ff7ab814adc1fd747f66fe059e6"}, "configs.json": {"size": 334579, "sha256": "4f04b994d73a5888e5f3d3c951ced24653db87dabf2cce779bba8bbeec1917aa"}, "provenance.json": {"size": 2485, "sha256": "4a4686297e4d6c4e7d204690e7bc30f5b12ac2a5784737db77db10ded980dbbb"}, "summary.json": {"size": 15793, "sha256": "7fd9177200c1ad25516a48c310e9877564e9b6a8349d3c064ceae891de64e03f"}, "arxivmath/configs.json": {"size": 118082, "sha256": "92eb191a9c75691f0cea623d4ddc67f2bee6f73b0f25a230e46451cf282d4311"}, "arxivmath/events.jsonl.gz": {"size": 154283104, "sha256": "04771ade3f1f9748a485efc8a57d62b267d8c7dfa61f28ce936ef2d3b12c0971"}, "arxivmath/judge_trajectories.jsonl": {"size": 1013339, "sha256": "b0587302ed141488f68a0150e0fd6f114accb4d7e0d27d1ce9f36e8104bfbc37"}, "arxivmath/provenance.json": {"size": 487, "sha256": "6cad9672125fcf583cd696c908afe1c5d3729347e97b41a65024f6970206c8ae"}, "arxivmath/summary.json": {"size": 6907, "sha256": "7f8d5fc68b4e63dfabf14dcaa9a218028eec3f70d7653eccf3c89b8914d1a796"}, "arxivmath/trajectories.jsonl": {"size": 251309308, "sha256": "f962b0b315f0cd78183cf07c4e7c9edb3d21205041b3eda074d8d3e15b82840c"}, "frontierscience-olympiad/configs.json": {"size": 118082, "sha256": "92eb191a9c75691f0cea623d4ddc67f2bee6f73b0f25a230e46451cf282d4311"}, "frontierscience-olympiad/events.jsonl.gz": {"size": 122060338, "sha256": "58773134bc4cdaa5970a5abbf98d1a41f85f438cfc175ebb04b52a4548fd57c1"}, "frontierscience-olympiad/judge_trajectories.jsonl": {"size": 10621950, "sha256": "fb8304c3a763723d0fb3cb5eeeab47aab14dff3a779ca9df45616d47f1691613"}, "frontierscience-olympiad/provenance.json": {"size": 487, "sha256": "6cad9672125fcf583cd696c908afe1c5d3729347e97b41a65024f6970206c8ae"}, "frontierscience-olympiad/summary.json": {"size": 1801, "sha256": "e50b6063426f1e215f42ecefb7a5e985da3cbe1b8727b09419af355bd7528b6c"}, "frontierscience-olympiad/trajectories.jsonl": {"size": 212047585, "sha256": "bfb05c3277c22845ac2565a29385f6c1e716a7050f1ae98557b302d7cd4595c8"}, "frontierscience-research/configs.json": {"size": 118082, "sha256": "92eb191a9c75691f0cea623d4ddc67f2bee6f73b0f25a230e46451cf282d4311"}, "frontierscience-research/events.jsonl.gz": {"size": 165019844, "sha256": "402309a7d8f519c690a48639e7b1c45c94e7f9c3bb8364b357a955fbb2493b7b"}, "frontierscience-research/judge_trajectories.jsonl": {"size": 28196794, "sha256": "3b9f3513e2a8a55dae2effb13e7a2f63dfcb5ee9be3da98a4ce2efa765f4512b"}, "frontierscience-research/provenance.json": {"size": 487, "sha256": "6cad9672125fcf583cd696c908afe1c5d3729347e97b41a65024f6970206c8ae"}, "frontierscience-research/summary.json": {"size": 1833, "sha256": "b2c89cbb7d1197aeda798b41f5390061f791efbfc57d78c1c61c006e28d9cf48"}, "frontierscience-research/trajectories.jsonl": {"size": 281994427, "sha256": "781b45c25feb6c62cff43684f06e5bd97ed0035b712dfce1b33cf30833e4bb76"}, "hle-verified/configs.json": {"size": 118082, "sha256": "92eb191a9c75691f0cea623d4ddc67f2bee6f73b0f25a230e46451cf282d4311"}, "hle-verified/events.jsonl.gz": {"size": 95588554, "sha256": "10f564a1b0f8b7df51635ca7e3ff4e905112a4295722b9e60b8a3a17d5e34f6a"}, "hle-verified/judge_trajectories.jsonl": {"size": 2398554, "sha256": "28c6718223faa79f2ad1b78df7e43f39b3001cca2e55f086c20b098240109143"}, "hle-verified/provenance.json": {"size": 295, "sha256": "abe752d84b30209d38d3569879dcd37c74b442ae083b53d5c0d87e26a41075aa"}, "hle-verified/summary.json": {"size": 3411, "sha256": "c532d14ef76215f7fa3f551d199ac0a1d173288a216433de3be1e684d397b371"}, "hle-verified/trajectories.jsonl": {"size": 159955220, "sha256": "8421efecf3eb9a9751a8c0f7f8a01d24ec63ef18aa9695fd4a344e1ade8573cf"}, "scicoding-agent/configs.json": {"size": 216470, "sha256": "d22d5d85c8ed58175c3af72dfed1d1f3dbd25a3fdd7a76455c2e1aa0de67864c"}, "scicoding-agent/judge_trajectories.jsonl": {"size": 1017721, "sha256": "bacee158c4e0dbfe642794d7eceb952f778b7c5fa2327ef69b32d728c2056ef3"}, "scicoding-agent/native_artifacts.tar.gz": {"size": 137140211, "sha256": "a8f7e99e6d5bff66a832a8c506621d4e345f64d49cc4e186cb422552d6e90a5d"}, "scicoding-agent/provenance.json": {"size": 369, "sha256": "1130676fba83089cadb9dffb748489cefb8710586b8ff7e58b175910e3e07e01"}, "scicoding-agent/summary.json": {"size": 566, "sha256": "7adb556362b146359a298db3bf642fbd127e17a4803c52f50b24dd88a4d9e70c"}, "scicoding-agent/trajectories.jsonl": {"size": 54443843, "sha256": "c0ed5bbb576568701cd98799f0e7d4229689f42ae4aec70a476bc99ad4c80868"}, "terminal-bench-2.1/configs.json": {"size": 216470, "sha256": "d22d5d85c8ed58175c3af72dfed1d1f3dbd25a3fdd7a76455c2e1aa0de67864c"}, "terminal-bench-2.1/judge_trajectories.jsonl": {"size": 6285607, "sha256": "0613850834b8344e260261bcb277507659ef4ec04a5c41dfabcf69681d8c3533"}, "terminal-bench-2.1/native_artifacts.tar.gz": {"size": 2845505047, "sha256": "4585dd0a66ba751b28e0ddc228a02a28d6de27e12ef6e76eeb2ad39c880f5e49"}, "terminal-bench-2.1/provenance.json": {"size": 369, "sha256": "1130676fba83089cadb9dffb748489cefb8710586b8ff7e58b175910e3e07e01"}, "terminal-bench-2.1/summary.json": {"size": 583, "sha256": "4d9631178228918c4ee322d24a7aad6605052954bbf5a1c6d8a5bb6e72cb2c65"}, "terminal-bench-2.1/trajectories.jsonl": {"size": 92359074, "sha256": "1fae7f6c2a2c4f831b5e33782963224f2ec21994c4404354988263b7fa89bc16"}}
|
rl-reasoning/intermediate/step-150-96k/provenance.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"agentic_publication": {"identity": {"repo": "JLiangHe/sci-ckpt-final", "revision": "a3d690d2356f3a8622f4b8047d2e98d23f3e9338", "folder": "rl-stage1-candidate/step-150-96k"}, "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116-agentic/01-rl-reasoning-step150-96k-full-20261116", "completed": {"time": 1794928290.2347572, "summary": {"overall": {"expected": 278, "completed": 278, "failed": 0, "pending": 0, "running_or_orphaned": 0, "passed": 69, "unscored": 0, "exceptions": 210, "finished": true, "tasks": 139, "attempts_per_task": 2, "metric": "avg@2", "mean_reward": 0.24820143884892087, "avg_at_2": 0.24820143884892087, "grading_complete": true, "provisional": false, "scoring_policy": "native-verifier-after-agent-stop-v1"}, "benchmarks": {"scicoding-agent": {"expected": 100, "completed": 100, "failed": 0, "pending": 0, "running_or_orphaned": 0, "passed": 24, "unscored": 0, "exceptions": 73, "finished": true, "tasks": 50, "attempts_per_task": 2, "metric": "avg@2", "mean_reward": 0.24, "avg_at_2": 0.24, "grading_complete": true, "provisional": false, "scoring_policy": "native-verifier-after-agent-stop-v1"}, "terminal-bench-2.1": {"expected": 178, "completed": 178, "failed": 0, "pending": 0, "running_or_orphaned": 0, "passed": 45, "unscored": 0, "exceptions": 137, "finished": true, "tasks": 89, "attempts_per_task": 2, "metric": "avg@2", "mean_reward": 0.25280898876404495, "avg_at_2": 0.25280898876404495, "grading_complete": true, "provisional": false, "scoring_policy": "native-verifier-after-agent-stop-v1"}}}, "execution_finished": true, "grading_complete": true, "unscored_audit": {"time": 1794928290.2273998, "execution_finished": true, "grading_complete": true, "provisional": false, "unscored": 0, "attempts": [], "policy": "Advance after all attempts terminate; retain unscored flags and original rewards; no automatic retry or zero assignment"}}, "regenerated_or_rejudged": false}, "reasoning_publication": {"source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116", "identity": {"repo": "JLiangHe/sci-ckpt-final", "revision": "a3d690d2356f3a8622f4b8047d2e98d23f3e9338", "folder": "rl-stage1-candidate/step-150-96k"}, "completed": {"time": 1794898203.0531983, "expected_trajectories": 8460, "scored": 8460, "results": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116/results"}, "published_date": "2026-11-17", "regenerated_or_rejudged": false}}
|
rl-reasoning/intermediate/step-150-96k/scicoding-agent/configs.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
rl-reasoning/intermediate/step-150-96k/scicoding-agent/judge_trajectories.jsonl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bacee158c4e0dbfe642794d7eceb952f778b7c5fa2327ef69b32d728c2056ef3
|
| 3 |
+
size 1017721
|
rl-reasoning/intermediate/step-150-96k/scicoding-agent/native_artifacts.tar.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a8f7e99e6d5bff66a832a8c506621d4e345f64d49cc4e186cb422552d6e90a5d
|
| 3 |
+
size 137140211
|
rl-reasoning/intermediate/step-150-96k/scicoding-agent/provenance.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"identity": {"repo": "JLiangHe/sci-ckpt-final", "revision": "a3d690d2356f3a8622f4b8047d2e98d23f3e9338", "folder": "rl-stage1-candidate/step-150-96k"}, "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116-agentic/01-rl-reasoning-step150-96k-full-20261116", "regenerated_or_rejudged": false, "manual_zero_adjudications": false}
|
rl-reasoning/intermediate/step-150-96k/scicoding-agent/summary.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"tasks": 50, "attempts": 100, "attempts_per_task": 2, "metric": "avg@2", "passed": 24, "avg_at_2": 0.24, "unscored": 0, "agent_timeouts": 0, "reward_positive_with_exception": 6, "scoring": "native_verifier", "scoring_policy": "native-verifier-after-agent-stop-v1", "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116-agentic/01-rl-reasoning-step150-96k-full-20261116", "provisional": false, "grading_complete": true, "sampling": {"temperature": 0.7, "top_p": 0.95, "top_k": 20}, "hardware": "H100", "tasks_per_engine": 2}
|
rl-reasoning/intermediate/step-150-96k/scicoding-agent/trajectories.jsonl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c0ed5bbb576568701cd98799f0e7d4229689f42ae4aec70a476bc99ad4c80868
|
| 3 |
+
size 54443843
|
rl-reasoning/intermediate/step-150-96k/summary.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"benchmarks": {"scicoding-agent": {"tasks": 50, "attempts": 100, "attempts_per_task": 2, "metric": "avg@2", "passed": 24, "avg_at_2": 0.24, "unscored": 0, "agent_timeouts": 0, "reward_positive_with_exception": 6, "scoring": "native_verifier", "scoring_policy": "native-verifier-after-agent-stop-v1", "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116-agentic/01-rl-reasoning-step150-96k-full-20261116", "provisional": false, "grading_complete": true, "sampling": {"temperature": 0.7, "top_p": 0.95, "top_k": 20}, "hardware": "H100", "tasks_per_engine": 2}, "terminal-bench-2.1": {"tasks": 89, "attempts": 178, "attempts_per_task": 2, "metric": "avg@2", "passed": 45, "avg_at_2": 0.25280898876404495, "unscored": 0, "agent_timeouts": 123, "reward_positive_with_exception": 8, "scoring": "native_verifier", "scoring_policy": "native-verifier-after-agent-stop-v1", "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116-agentic/01-rl-reasoning-step150-96k-full-20261116", "provisional": false, "grading_complete": true, "sampling": {"temperature": 0.7, "top_p": 0.95, "top_k": 20}, "hardware": "H100", "tasks_per_engine": 2}, "frontierscience-olympiad": {"total_records": 3000, "completed": 3000, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.6443333333333333, "accuracy": 0.6443333333333333, "mean_latency_seconds": 302.8745814699524, "finish_reason_counts": {"length": 4, "stop": 2996}, "length_limited_response_count": 4, "observed_max_completion_tokens": 98304, "judge_usage": {"judged_record_count": 3000, "request_count": 3000, "retry_request_count": 0, "prompt_tokens": 2571739, "cached_prompt_tokens": 1278, "cache_write_prompt_tokens": 762084, "completion_tokens": 839197, "reasoning_tokens": 612675, "total_tokens": 3410936}, "judge_api_usage": {"request_count": 3000, "prompt_tokens": 2571739, "cached_prompt_tokens": 1278, "cache_write_prompt_tokens": 762084, "completion_tokens": 839197, "reasoning_tokens": 612675, "total_tokens": 3410936}, "benchmark_metrics": {"track": "olympiad", "n_problems_scored": 100, "n_trajectories_scored": 3000, "aggregation": "avg@30", "reported_accuracy": 0.6443333333333334, "mean_rubric_score": null, "success_threshold": 0.5, "community_profile": null}, "benchmark": "frontierscience-olympiad", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "e7cd88d9f221c8384c641ecff9ac33e42425a247bbff06960324d88f8208b266", "scoring_hash": "0e899897c7cf45f80e15edc01a869181d6f588ca19d5dc3bcb855d8325f66209", "selected": 3000, "incomplete": 0, "generated_now": 3000, "scored_now": 3000, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 3000, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T05:32:02.376642+00:00", "unscored": 0, "provisional": false, "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116"}, "frontierscience-research": {"total_records": 1800, "completed": 1800, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.34740944444444444, "accuracy": 0.1411111111111111, "mean_latency_seconds": 538.5143223827538, "finish_reason_counts": {"length": 8, "stop": 1792}, "length_limited_response_count": 8, "observed_max_completion_tokens": 98304, "judge_usage": {"judged_record_count": 1800, "request_count": 1831, "retry_request_count": 31, "prompt_tokens": 13111413, "cached_prompt_tokens": 259361, "cache_write_prompt_tokens": 12836474, "completion_tokens": 3156537, "reasoning_tokens": 1893337, "total_tokens": 16267950}, "judge_api_usage": {"request_count": 1831, "prompt_tokens": 13111413, "cached_prompt_tokens": 259361, "cache_write_prompt_tokens": 12836474, "completion_tokens": 3156537, "reasoning_tokens": 1893337, "total_tokens": 16267950}, "benchmark_metrics": {"track": "research", "n_problems_scored": 60, "n_trajectories_scored": 1800, "aggregation": "avg@30", "reported_accuracy": 0.1411111111111111, "mean_rubric_score": 0.34740944444444444, "success_threshold": 0.7, "community_profile": null}, "benchmark": "frontierscience-research", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "f49bda87d537a6fd83c21f5ff513cc67c58d8701715f216d50e640c334c0dde8", "scoring_hash": "e94267060b6ed01e455b1c25c69625896aaa44778fa60f6eddc9d577e40412b5", "selected": 1800, "incomplete": 0, "generated_now": 1800, "scored_now": 1800, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 1800, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T04:30:00.577547+00:00", "unscored": 0, "provisional": false, "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116"}, "arxivmath": {"benchmark": "arxivmath", "releases": [{"total_records": 340, "completed": 340, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.36176470588235293, "accuracy": 0.36176470588235293, "mean_latency_seconds": 670.1104181452298, "finish_reason_counts": {"length": 4, "stop": 336}, "length_limited_response_count": 4, "observed_max_completion_tokens": 98304, "judge_usage": {"judged_record_count": 0, "request_count": 0, "retry_request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "judge_api_usage": {"request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "benchmark_metrics": {"competition": "arxivmath", "aggregation": "avg@20", "n_problems_scored": 17, "n_trajectories_scored": 340, "reported_accuracy": 0.361764705882353, "competition_metrics": {"arxiv/december": {"n_problems_scored": 17, "n_trajectories_scored": 340, "reported_accuracy": 0.361764705882353}}}, "benchmark": "matharena-arxivmath-december", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "337d6945d81da4a84ccffc6620608096d7b562a55cd1ccd41017e23ce77061e1", "scoring_hash": "4d4e5f2b4313ddf081bf1c70a32baf7f40b9f05b9fd6d60158bcf12d7f43b1ae", "selected": 340, "incomplete": 0, "generated_now": 340, "scored_now": 340, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 340, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T06:15:24.505576+00:00"}, {"total_records": 640, "completed": 640, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.234375, "accuracy": 0.234375, "mean_latency_seconds": 571.4612996978431, "finish_reason_counts": {"stop": 640}, "length_limited_response_count": 0, "observed_max_completion_tokens": 93642, "judge_usage": {"judged_record_count": 0, "request_count": 0, "retry_request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "judge_api_usage": {"request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "benchmark_metrics": {"competition": "arxivmath", "aggregation": "avg@20", "n_problems_scored": 32, "n_trajectories_scored": 640, "reported_accuracy": 0.234375, "competition_metrics": {"arxiv/february": {"n_problems_scored": 32, "n_trajectories_scored": 640, "reported_accuracy": 0.234375}}}, "benchmark": "matharena-arxivmath-february", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "051ccdbf7a0d3dba10c0e849d0a12a0d1e6cd2801da566575204f75dc7cfe92b", "scoring_hash": "821b7ad517ea717cb02ac5d1093152349dd9881b04b883d11872c23b4e63cfef", "selected": 640, "incomplete": 0, "generated_now": 640, "scored_now": 640, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 640, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T05:58:50.360285+00:00"}, {"total_records": 460, "completed": 460, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.45652173913043476, "accuracy": 0.45652173913043476, "mean_latency_seconds": 727.2866143759867, "finish_reason_counts": {"length": 3, "stop": 457}, "length_limited_response_count": 3, "observed_max_completion_tokens": 98304, "judge_usage": {"judged_record_count": 0, "request_count": 0, "retry_request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "judge_api_usage": {"request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "benchmark_metrics": {"competition": "arxivmath", "aggregation": "avg@20", "n_problems_scored": 23, "n_trajectories_scored": 460, "reported_accuracy": 0.45652173913043476, "competition_metrics": {"arxiv/january": {"n_problems_scored": 23, "n_trajectories_scored": 460, "reported_accuracy": 0.45652173913043476}}}, "benchmark": "matharena-arxivmath-january", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "e968177909e2e4b539e069951ebae4b7515e5b753d9136afac8e6a4f18fc9d02", "scoring_hash": "09da72fc33a67c50f76c5013fce57317c2b3ddc2989f723fbc906ef5a8941e69", "selected": 460, "incomplete": 0, "generated_now": 460, "scored_now": 460, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 460, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T06:40:55.271786+00:00"}, {"total_records": 620, "completed": 620, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.23225806451612904, "accuracy": 0.23225806451612904, "mean_latency_seconds": 513.2421608521363, "finish_reason_counts": {"stop": 620}, "length_limited_response_count": 0, "observed_max_completion_tokens": 83370, "judge_usage": {"judged_record_count": 0, "request_count": 0, "retry_request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "judge_api_usage": {"request_count": 0, "prompt_tokens": 0, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 0, "completion_tokens": 0, "reasoning_tokens": 0, "total_tokens": 0}, "benchmark_metrics": {"competition": "arxivmath", "aggregation": "avg@20", "n_problems_scored": 31, "n_trajectories_scored": 620, "reported_accuracy": 0.23225806451612904, "competition_metrics": {"arxiv/march": {"n_problems_scored": 31, "n_trajectories_scored": 620, "reported_accuracy": 0.23225806451612904}}}, "benchmark": "matharena-arxivmath-march", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "c94c0faa3d44df057bc05f7731d866cbef8c9e251af4802054bee5e9b503557e", "scoring_hash": "34ddbb860c27cc076bdb3606863c8ffe8447880958ca9974cd1c6dcc6bab7543", "selected": 620, "incomplete": 0, "generated_now": 620, "scored_now": 620, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 620, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T06:49:54.272195+00:00"}], "total_records": 2060, "completed": 2060, "accuracy": 0.30436893203883497, "correct": 627.0, "aggregation": "avg@20", "trials": 20, "questions": 103, "unscored": 0, "provisional": false, "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116"}, "hle-verified": {"benchmark": "hle-verified", "total_records": 1600, "completed": 1600, "correct": 511.0, "accuracy": 0.319375, "mean_score": 0.319375, "aggregation": "question-weighted Gold + Revision, one trial per question", "subsets": {"gold": {"total_records": 575, "completed": 575, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.2956521739130435, "accuracy": 0.2956521739130435, "mean_latency_seconds": 460.71861102186995, "finish_reason_counts": {"stop": 575}, "length_limited_response_count": 0, "observed_max_completion_tokens": 94781, "judge_usage": {"judged_record_count": 575, "request_count": 575, "retry_request_count": 0, "prompt_tokens": 395254, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 91416, "completion_tokens": 73538, "reasoning_tokens": 39228, "total_tokens": 468792}, "judge_api_usage": {"request_count": 575, "prompt_tokens": 395254, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 91416, "completion_tokens": 73538, "reasoning_tokens": 39228, "total_tokens": 468792}, "benchmark_metrics": {"accuracy": 0.2956521739130435, "calibration_error": 58.41285092873989, "calibration_minimum_n": 100}, "benchmark": "hle-verified-gold", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "7140f7ed7115665f0a2617a4d301fd854fe5bf6312d72b86c3f06f2bec7aa62c", "scoring_hash": "546d1fcbe7376d61b7ca0a6637261ff33ba90afc425bb4b4d5219c1da57b3645", "selected": 575, "incomplete": 0, "generated_now": 575, "scored_now": 575, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 575, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T03:38:05.567376+00:00"}, "revision": {"total_records": 1025, "completed": 1025, "query_errors": 0, "score_errors": 0, "generated_unscored": 0, "mean_score": 0.3326829268292683, "accuracy": 0.3326829268292683, "mean_latency_seconds": 408.6073815469567, "finish_reason_counts": {"length": 4, "stop": 1021}, "length_limited_response_count": 4, "observed_max_completion_tokens": 98304, "judge_usage": {"judged_record_count": 1025, "request_count": 1025, "retry_request_count": 0, "prompt_tokens": 750729, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 189092, "completion_tokens": 143073, "reasoning_tokens": 79454, "total_tokens": 893802}, "judge_api_usage": {"request_count": 1025, "prompt_tokens": 750729, "cached_prompt_tokens": 0, "cache_write_prompt_tokens": 189092, "completion_tokens": 143073, "reasoning_tokens": 79454, "total_tokens": 893802}, "benchmark_metrics": {"accuracy": 0.3326829268292683, "calibration_error": 58.53045994577024, "calibration_minimum_n": 100}, "benchmark": "hle-verified-revision", "model": "rl-reasoning-step150-96k-reasoning", "execution_mode": "non_agent", "generation_hash": "9af86219049738844a6f1c0036cef9232d8d6df323e26e3e80f35d28967bf21c", "scoring_hash": "4766321e2412a75c642ebeabbf674dcd2f8409d2e5148e2e8a4a46f7619d9073", "selected": 1025, "incomplete": 0, "generated_now": 1025, "scored_now": 1025, "cached_completed": 0, "stage": "all", "worker_count": 1, "worker_index": 0, "worker_owned_predictions": 1025, "generation_concurrency": 256, "judge_concurrency": 256, "updated_at": "2026-11-17T03:33:02.556528+00:00"}}, "unscored": 0, "provisional": false, "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116"}}, "unscored": {"scicoding-agent": 0, "terminal-bench-2.1": 0, "frontierscience-olympiad": 0, "frontierscience-research": 0, "arxivmath": 0, "hle-verified": 0}, "checkpoint": "/workspace/data/raw_model/sci-ckpt-final/a3d690d2356f3a8622f4b8047d2e98d23f3e9338/rl-stage1-candidate/step-150-96k", "identity": {"repo": "JLiangHe/sci-ckpt-final", "revision": "a3d690d2356f3a8622f4b8047d2e98d23f3e9338", "folder": "rl-stage1-candidate/step-150-96k"}, "generation_finished": true, "fully_scored": true, "full_suite_complete": true, "provisional": false}
|
rl-reasoning/intermediate/step-150-96k/terminal-bench-2.1/configs.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
rl-reasoning/intermediate/step-150-96k/terminal-bench-2.1/judge_trajectories.jsonl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0613850834b8344e260261bcb277507659ef4ec04a5c41dfabcf69681d8c3533
|
| 3 |
+
size 6285607
|
rl-reasoning/intermediate/step-150-96k/terminal-bench-2.1/native_artifacts.tar.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4585dd0a66ba751b28e0ddc228a02a28d6de27e12ef6e76eeb2ad39c880f5e49
|
| 3 |
+
size 2845505047
|
rl-reasoning/intermediate/step-150-96k/terminal-bench-2.1/provenance.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"identity": {"repo": "JLiangHe/sci-ckpt-final", "revision": "a3d690d2356f3a8622f4b8047d2e98d23f3e9338", "folder": "rl-stage1-candidate/step-150-96k"}, "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116-agentic/01-rl-reasoning-step150-96k-full-20261116", "regenerated_or_rejudged": false, "manual_zero_adjudications": false}
|
rl-reasoning/intermediate/step-150-96k/terminal-bench-2.1/summary.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"tasks": 89, "attempts": 178, "attempts_per_task": 2, "metric": "avg@2", "passed": 45, "avg_at_2": 0.25280898876404495, "unscored": 0, "agent_timeouts": 123, "reward_positive_with_exception": 8, "scoring": "native_verifier", "scoring_policy": "native-verifier-after-agent-stop-v1", "source_run": "/workspace/data/sci-llm-training/runtime/rl-reasoning-step150-96k-full-20261116-agentic/01-rl-reasoning-step150-96k-full-20261116", "provisional": false, "grading_complete": true, "sampling": {"temperature": 0.7, "top_p": 0.95, "top_k": 20}, "hardware": "H100", "tasks_per_engine": 2}
|
rl-reasoning/intermediate/step-150-96k/terminal-bench-2.1/trajectories.jsonl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1fae7f6c2a2c4f831b5e33782963224f2ec21994c4404354988263b7fa89bc16
|
| 3 |
+
size 92359074
|