Datasets:
Release v1.0.1: trim text mix to ~50%, add stage-2 checkpoint and agent validation/test splits
Browse filesRemoved four complete public-domain literary sources (8,285 tasks), kept the main task schema stable, published 272 anonymised stage-2 contexts as stage2/added_2026-11-13.jsonl, added 90/5/5 splits for agent trajectories and steps, and rebuilt reports, manifest and README.
- README.md +48 -46
- agents/agent_steps.jsonl +0 -0
- agents/agent_trajectories.jsonl +0 -0
- capture/compact.jsonl +2 -2
- capture/extended.jsonl +2 -2
- data/test.jsonl +2 -2
- data/train.jsonl +2 -2
- data/validation.jsonl +2 -2
- manifest.json +26 -24
- privacy_report.json +3 -8
- quality_report.json +122 -110
- stage2/added_2026-11-13.jsonl +0 -0
README.md
CHANGED
|
@@ -729,7 +729,7 @@ dataset_info:
|
|
| 729 |
|
| 730 |
**DeepSeek teacher traces. Contexts for the next draft.**
|
| 731 |
|
| 732 |
-
`41,
|
| 733 |
|
| 734 |
[Explore the data](https://huggingface.co/datasets/biscet/DFlash-DeepSeek-TraceMix/tree/main) · [Manifest](https://huggingface.co/datasets/biscet/DFlash-DeepSeek-TraceMix/blob/main/manifest.json) · [Privacy report](https://huggingface.co/datasets/biscet/DFlash-DeepSeek-TraceMix/blob/main/privacy_report.json)
|
| 735 |
|
|
@@ -741,7 +741,7 @@ dataset_info:
|
|
| 741 |
> A bilingual mix of task prompts and DeepSeek-generated agent traces for
|
| 742 |
> preparing draft-model training and studying tool use.
|
| 743 |
|
| 744 |
-
The main corpus contains **41,
|
| 745 |
completions. The separate agent supplements contain actual reasoning, historical
|
| 746 |
tool calls, real observations and final responses distilled from the DeepSeek
|
| 747 |
teacher models listed below through Qwen Code.
|
|
@@ -750,72 +750,75 @@ teacher models listed below through Qwen Code.
|
|
| 750 |
|
| 751 |
- **Text corpus trimmed.** Four complete public-domain literary sources were
|
| 752 |
removed (8,285 tasks in total) to bring the text share down to about
|
| 753 |
-
one half of the corpus.
|
| 754 |
-
- **New stage-2
|
| 755 |
(`project_03`, `project_05_mobile`) were converted to the release schema,
|
| 756 |
-
anonymised and
|
| 757 |
-
|
| 758 |
-
|
|
|
|
| 759 |
- **Splits rebalanced** proportionally to 90 / 5 / 5.
|
| 760 |
|
| 761 |
## Stage-2 checkpoint (2026-11-13)
|
| 762 |
|
| 763 |
-
The
|
| 764 |
checkpoint, so they can be pulled out of the full dataset without filtering:
|
| 765 |
|
| 766 |
[`stage2/added_2026-11-13.jsonl`](https://huggingface.co/datasets/biscet/DFlash-DeepSeek-TraceMix/blob/main/stage2/added_2026-11-13.jsonl)
|
| 767 |
|
| 768 |
-
- **272 records**, one JSON object per line,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 769 |
- Every record carries `provenance.generator_version = "1.0.1+stage2-2026-11-13"`.
|
| 770 |
- `source_sha256` is computed over the anonymised message/tool context, not over
|
| 771 |
the original repository file.
|
| 772 |
-
- They are the same objects that appear in the task splits (190 train,
|
| 773 |
-
42 validation, 40 test), so loading both would duplicate
|
| 774 |
-
them; use the checkpoint for extraction and `data/*.jsonl` for the complete
|
| 775 |
-
training mix.
|
| 776 |
|
| 777 |
## Teacher models
|
| 778 |
|
| 779 |
We distilled the agent traces from **DeepSeek Flash v4.1 (1017)** and
|
| 780 |
**DeepSeek Pro V4 (0813)**.
|
| 781 |
|
| 782 |
-
| Teacher model | Recorded API alias |
|
| 783 |
-
|:---|:---|---:|
|
| 784 |
-
| DeepSeek Flash v4.1 (1017) | `deepseek-flash` | 161 |
|
| 785 |
-
| DeepSeek Pro V4 (0813) | `deepseek-v4-pro` | 39 |
|
| 786 |
|
|
|
|
|
|
|
| 787 |
Trace records retain the original API aliases in their model and provenance fields.
|
| 788 |
|
| 789 |
## Repository branches
|
| 790 |
|
| 791 |
| Branch | Contents |
|
| 792 |
|:---|:---|
|
| 793 |
-
| `main` |
|
| 794 |
-
| [`v1.0.1`](https://huggingface.co/datasets/biscet/DFlash-DeepSeek-TraceMix/tree/v1.0.1) | Archived v1.0.1 release snapshot: trimmed text mix and stage-2 checkpoint. |
|
| 795 |
| [`v1.0.0`](https://huggingface.co/datasets/biscet/DFlash-DeepSeek-TraceMix/tree/v1.0.0) | Archived v1.0.0 release snapshot with the full 50,000-task text mix. |
|
| 796 |
|
| 797 |
The complete dataset is available from the default `main` branch and in the
|
| 798 |
-
Dataset Viewer.
|
| 799 |
-
`revision="v1.0.
|
| 800 |
-
|
| 801 |
-
evaluation splits.
|
| 802 |
|
| 803 |
## The mix
|
| 804 |
|
| 805 |
| Configuration | Focus | Train | Validation | Test |
|
| 806 |
|:---|:---|---:|---:|---:|
|
| 807 |
-
| `tasks` | Coding, text processing, knowledge and reasoning contexts | 37,
|
| 808 |
-
| `agent_steps` | One next-action step with tool history and observations |
|
| 809 |
-
| `agent_trajectories` | Complete DeepSeek sessions with reasoning and tool use |
|
|
|
|
| 810 |
|
| 811 |
The 294 steps are derived from the 200 trajectories. They are two views of the
|
| 812 |
same agent sessions, not independent extra sessions.
|
| 813 |
|
| 814 |
-
- **Main languages:** 20,
|
| 815 |
-
- **Task mix:** 52.
|
| 816 |
-
- **Source mix:** 21,313 literary-text tasks, 19,
|
| 817 |
1,262 upstream prompt tasks and 58 self-created tasks.
|
| 818 |
-
- **Splits:** 10,
|
| 819 |
- **Agent traces:** 604 reasoning blocks and 611 tool calls across 200 teacher
|
| 820 |
sessions, kept separate from the task contexts.
|
| 821 |
|
|
@@ -824,12 +827,12 @@ same agent sessions, not independent extra sessions.
|
|
| 824 |
| Path | Contents |
|
| 825 |
|:---|:---|
|
| 826 |
| `data/train.jsonl`, `data/validation.jsonl`, `data/test.jsonl` | Task IDs, messages, source/provenance, rendered prompts and exact token lengths. |
|
| 827 |
-
| `stage2/added_2026-11-13.jsonl` | The 272
|
| 828 |
| `agents/agent_trajectories.jsonl` | Sanitized complete DeepSeek trajectories. |
|
| 829 |
| `agents/agent_steps.jsonl` | Sanitized next-action records compatible with the task schema. |
|
| 830 |
| `capture/compact.jsonl` | Whittle capture inputs, at most 1,536 prompt tokens. |
|
| 831 |
| `capture/extended.jsonl` | Longer Whittle capture inputs, 1,537–8,192 prompt tokens. |
|
| 832 |
-
| `quality_report.json` | Main-corpus schema, split, duplicate, token-length and
|
| 833 |
| `privacy_report.json` | Publication cleanup scope and automated privacy checks. |
|
| 834 |
| `manifest.json` | Release counts, file sizes and SHA-256 checksums. |
|
| 835 |
| `LICENSE`, `NOTICE` | Apache-2.0 release license and retained source attribution. |
|
|
@@ -845,7 +848,7 @@ tasks = load_dataset(
|
|
| 845 |
agent_steps = load_dataset(
|
| 846 |
"biscet/DFlash-DeepSeek-TraceMix", "agent_steps"
|
| 847 |
)
|
| 848 |
-
# v1.0.1
|
| 849 |
stage2 = load_dataset(
|
| 850 |
"biscet/DFlash-DeepSeek-TraceMix",
|
| 851 |
data_files="stage2/added_2026-11-13.jsonl",
|
|
@@ -864,16 +867,15 @@ stage2 = load_dataset(
|
|
| 864 |
logits or captured teacher token IDs. Generate the target model's responses
|
| 865 |
and features through your capture pipeline before draft-model training.
|
| 866 |
- **Teacher scope:** DeepSeek provenance applies to the agent configurations.
|
| 867 |
-
The task-context
|
| 868 |
-
- **Lengths:** main-task prompts range from 52 to
|
| 869 |
-
|
| 870 |
-
|
| 871 |
-
|
| 872 |
-
separately.
|
| 873 |
- **Quality:** the saved main-corpus checks pass, with no detected exact or near
|
| 874 |
-
duplicates across families
|
| 875 |
-
|
| 876 |
-
not verified reference answers or a held-out benchmark.
|
| 877 |
|
| 878 |
</details>
|
| 879 |
|
|
@@ -890,10 +892,10 @@ project names were replaced with stable placeholders, repository file paths were
|
|
| 890 |
replaced with hashed names, and commit revisions were dropped. Changed task
|
| 891 |
prompts and token lengths were recalculated with the Whittle tokenizer.
|
| 892 |
|
| 893 |
-
The publication audit
|
| 894 |
-
copies, and found no remaining matches to the configured
|
| 895 |
-
This is an automated pattern audit, not a guarantee that
|
| 896 |
-
identifying detail was removed.
|
| 897 |
|
| 898 |
Raw logs, local configuration, checkpoints, reference solutions and rejected
|
| 899 |
candidates are excluded. See the [privacy report](https://huggingface.co/datasets/biscet/DFlash-DeepSeek-TraceMix/blob/main/privacy_report.json).
|
|
|
|
| 729 |
|
| 730 |
**DeepSeek teacher traces. Contexts for the next draft.**
|
| 731 |
|
| 732 |
+
`41,715 task contexts` · `272 stage-2 contexts` · `200 agent trajectories` · `294 next-action steps` · `RU / EN`
|
| 733 |
|
| 734 |
[Explore the data](https://huggingface.co/datasets/biscet/DFlash-DeepSeek-TraceMix/tree/main) · [Manifest](https://huggingface.co/datasets/biscet/DFlash-DeepSeek-TraceMix/blob/main/manifest.json) · [Privacy report](https://huggingface.co/datasets/biscet/DFlash-DeepSeek-TraceMix/blob/main/privacy_report.json)
|
| 735 |
|
|
|
|
| 741 |
> A bilingual mix of task prompts and DeepSeek-generated agent traces for
|
| 742 |
> preparing draft-model training and studying tool use.
|
| 743 |
|
| 744 |
+
The main corpus contains **41,715 prompts and contexts**, not DeepSeek
|
| 745 |
completions. The separate agent supplements contain actual reasoning, historical
|
| 746 |
tool calls, real observations and final responses distilled from the DeepSeek
|
| 747 |
teacher models listed below through Qwen Code.
|
|
|
|
| 750 |
|
| 751 |
- **Text corpus trimmed.** Four complete public-domain literary sources were
|
| 752 |
removed (8,285 tasks in total) to bring the text share down to about
|
| 753 |
+
one half of the main corpus.
|
| 754 |
+
- **New stage-2 contexts added.** 272 tasks built from two real codebases
|
| 755 |
(`project_03`, `project_05_mobile`) were converted to the release schema,
|
| 756 |
+
anonymised and published as a standalone stage-2 checkpoint for the
|
| 757 |
+
**DeepSeek Flash** capture track.
|
| 758 |
+
- **Resulting main mix:** 52.6% text/reasoning and 47.4%
|
| 759 |
+
software development, with 20,546 Russian and 21,169 English tasks.
|
| 760 |
- **Splits rebalanced** proportionally to 90 / 5 / 5.
|
| 761 |
|
| 762 |
## Stage-2 checkpoint (2026-11-13)
|
| 763 |
|
| 764 |
+
The 272 contexts added in this release are published as a standalone
|
| 765 |
checkpoint, so they can be pulled out of the full dataset without filtering:
|
| 766 |
|
| 767 |
[`stage2/added_2026-11-13.jsonl`](https://huggingface.co/datasets/biscet/DFlash-DeepSeek-TraceMix/blob/main/stage2/added_2026-11-13.jsonl)
|
| 768 |
|
| 769 |
+
- **272 records**, one JSON object per line, `split = "train"` in the file.
|
| 770 |
+
- These records are **not** duplicated inside `data/*.jsonl`; the main `tasks`
|
| 771 |
+
configuration keeps its original schema and split boundaries.
|
| 772 |
+
- They are prompt contexts prepared for the DeepSeek Flash capture stage, not
|
| 773 |
+
completed DeepSeek trajectories, so they are not added to the teacher
|
| 774 |
+
trajectory counts.
|
| 775 |
- Every record carries `provenance.generator_version = "1.0.1+stage2-2026-11-13"`.
|
| 776 |
- `source_sha256` is computed over the anonymised message/tool context, not over
|
| 777 |
the original repository file.
|
|
|
|
|
|
|
|
|
|
|
|
|
| 778 |
|
| 779 |
## Teacher models
|
| 780 |
|
| 781 |
We distilled the agent traces from **DeepSeek Flash v4.1 (1017)** and
|
| 782 |
**DeepSeek Pro V4 (0813)**.
|
| 783 |
|
| 784 |
+
| Teacher model | Recorded API alias | Completed trajectories | v1.0.1 task contexts |
|
| 785 |
+
|:---|:---|---:|---:|
|
| 786 |
+
| DeepSeek Flash v4.1 (1017) | `deepseek-flash` | 161 | 272 |
|
| 787 |
+
| DeepSeek Pro V4 (0813) | `deepseek-v4-pro` | 39 | — |
|
| 788 |
|
| 789 |
+
The 272 new contexts are inputs for the DeepSeek Flash capture track;
|
| 790 |
+
they are not completed sessions, so the trajectory counts stay unchanged.
|
| 791 |
Trace records retain the original API aliases in their model and provenance fields.
|
| 792 |
|
| 793 |
## Repository branches
|
| 794 |
|
| 795 |
| Branch | Contents |
|
| 796 |
|:---|:---|
|
| 797 |
+
| `main` | Current **v1.0.1** release: trimmed task contexts, stage-2 checkpoint, DeepSeek agent data, capture exports, manifests and reports. |
|
|
|
|
| 798 |
| [`v1.0.0`](https://huggingface.co/datasets/biscet/DFlash-DeepSeek-TraceMix/tree/v1.0.0) | Archived v1.0.0 release snapshot with the full 50,000-task text mix. |
|
| 799 |
|
| 800 |
The complete dataset is available from the default `main` branch and in the
|
| 801 |
+
Dataset Viewer. `v1.0.0` preserves the previous release snapshot; pass
|
| 802 |
+
`revision="v1.0.0"` for it. Agent records follow the same 90/5/5 split policy
|
| 803 |
+
by trajectory family and stay separate from the task splits.
|
|
|
|
| 804 |
|
| 805 |
## The mix
|
| 806 |
|
| 807 |
| Configuration | Focus | Train | Validation | Test |
|
| 808 |
|:---|:---|---:|---:|---:|
|
| 809 |
+
| `tasks` | Coding, text processing, knowledge and reasoning contexts | 37,500 | 2,110 | 2,105 |
|
| 810 |
+
| `agent_steps` | One next-action step with tool history and observations | 268 | 14 | 12 |
|
| 811 |
+
| `agent_trajectories` | Complete DeepSeek sessions with reasoning and tool use | 180 | 10 | 10 |
|
| 812 |
+
| `stage2` (file) | Stage-2 contexts for the DeepSeek Flash capture track | 272 | — | — |
|
| 813 |
|
| 814 |
The 294 steps are derived from the 200 trajectories. They are two views of the
|
| 815 |
same agent sessions, not independent extra sessions.
|
| 816 |
|
| 817 |
+
- **Main languages:** 20,546 Russian and 21,169 English tasks; no machine translations.
|
| 818 |
+
- **Task mix:** 52.6% text/reasoning tasks (21,922) and 47.4% software development (19,793).
|
| 819 |
+
- **Source mix:** 21,313 literary-text tasks, 19,082 anonymized project-code tasks,
|
| 820 |
1,262 upstream prompt tasks and 58 self-created tasks.
|
| 821 |
+
- **Splits:** 10,076 semantic families, with each family confined to one split.
|
| 822 |
- **Agent traces:** 604 reasoning blocks and 611 tool calls across 200 teacher
|
| 823 |
sessions, kept separate from the task contexts.
|
| 824 |
|
|
|
|
| 827 |
| Path | Contents |
|
| 828 |
|:---|:---|
|
| 829 |
| `data/train.jsonl`, `data/validation.jsonl`, `data/test.jsonl` | Task IDs, messages, source/provenance, rendered prompts and exact token lengths. |
|
| 830 |
+
| `stage2/added_2026-11-13.jsonl` | The 272 stage-2 contexts added in v1.0.1, split `train`. |
|
| 831 |
| `agents/agent_trajectories.jsonl` | Sanitized complete DeepSeek trajectories. |
|
| 832 |
| `agents/agent_steps.jsonl` | Sanitized next-action records compatible with the task schema. |
|
| 833 |
| `capture/compact.jsonl` | Whittle capture inputs, at most 1,536 prompt tokens. |
|
| 834 |
| `capture/extended.jsonl` | Longer Whittle capture inputs, 1,537–8,192 prompt tokens. |
|
| 835 |
+
| `quality_report.json` | Main-corpus schema, split, duplicate, token-length and privacy checks. |
|
| 836 |
| `privacy_report.json` | Publication cleanup scope and automated privacy checks. |
|
| 837 |
| `manifest.json` | Release counts, file sizes and SHA-256 checksums. |
|
| 838 |
| `LICENSE`, `NOTICE` | Apache-2.0 release license and retained source attribution. |
|
|
|
|
| 848 |
agent_steps = load_dataset(
|
| 849 |
"biscet/DFlash-DeepSeek-TraceMix", "agent_steps"
|
| 850 |
)
|
| 851 |
+
# v1.0.1 stage-2 checkpoint
|
| 852 |
stage2 = load_dataset(
|
| 853 |
"biscet/DFlash-DeepSeek-TraceMix",
|
| 854 |
data_files="stage2/added_2026-11-13.jsonl",
|
|
|
|
| 867 |
logits or captured teacher token IDs. Generate the target model's responses
|
| 868 |
and features through your capture pipeline before draft-model training.
|
| 869 |
- **Teacher scope:** DeepSeek provenance applies to the agent configurations.
|
| 870 |
+
The task-context configurations are not corpora of DeepSeek completions.
|
| 871 |
+
- **Lengths:** main-task prompts range from 52 to 5,030 tokens.
|
| 872 |
+
Stage-2 contexts are longer and are not part of `capture/`; the largest
|
| 873 |
+
stage-2 prompt is 12,967 tokens. Agent steps range from 1,128 to 8,176 tokens.
|
| 874 |
+
Run compact and extended capture separately.
|
|
|
|
| 875 |
- **Quality:** the saved main-corpus checks pass, with no detected exact or near
|
| 876 |
+
duplicates across families. Stage-2 contexts are a separate supplement and
|
| 877 |
+
should be evaluated on their own split. Agent outputs are model-generated
|
| 878 |
+
material, not verified reference answers or a held-out benchmark.
|
| 879 |
|
| 880 |
</details>
|
| 881 |
|
|
|
|
| 892 |
replaced with hashed names, and commit revisions were dropped. Changed task
|
| 893 |
prompts and token lengths were recalculated with the Whittle tokenizer.
|
| 894 |
|
| 895 |
+
The publication audit covered the task rows, the stage-2 checkpoint, the agent
|
| 896 |
+
records and the capture copies, and found no remaining matches to the configured
|
| 897 |
+
sensitive-data patterns. This is an automated pattern audit, not a guarantee that
|
| 898 |
+
every possible identifying detail was removed.
|
| 899 |
|
| 900 |
Raw logs, local configuration, checkpoints, reference solutions and rejected
|
| 901 |
candidates are excluded. See the [privacy report](https://huggingface.co/datasets/biscet/DFlash-DeepSeek-TraceMix/blob/main/privacy_report.json).
|
agents/agent_steps.jsonl
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
agents/agent_trajectories.jsonl
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
capture/compact.jsonl
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9124c78a080ec0856d4e75ab24c53b7a3a231cb92adb4b95e8efe94159b232ae
|
| 3 |
+
size 115627119
|
capture/extended.jsonl
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:56dc5dcd4486ed159c0ceff1994a6de4433731c4b4673d3ec0e8a915e1b780f5
|
| 3 |
+
size 10162310
|
data/test.jsonl
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:051024f1fc0ff64eb1fb79d2012d7f08dc7459f77737d04ea4b27730786c7182
|
| 3 |
+
size 12497371
|
data/train.jsonl
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c7b886614647c1f67a8b0db45d2b34f6af875e599e4318118af9e7c1c97a1569
|
| 3 |
+
size 265345388
|
data/validation.jsonl
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5b6e6eb5fa0b068f049ffc53510ec69b297e5d3b2e07257e2b410a38dbd8a78e
|
| 3 |
+
size 13349115
|
manifest.json
CHANGED
|
@@ -1,5 +1,7 @@
|
|
| 1 |
{
|
| 2 |
-
"tasks":
|
|
|
|
|
|
|
| 3 |
"trajectories": 200,
|
| 4 |
"steps": 294,
|
| 5 |
"files": {
|
|
@@ -16,52 +18,52 @@
|
|
| 16 |
"bytes": 1689
|
| 17 |
},
|
| 18 |
"README.md": {
|
| 19 |
-
"sha256": "
|
| 20 |
-
"bytes":
|
| 21 |
},
|
| 22 |
"agents/agent_steps.jsonl": {
|
| 23 |
-
"sha256": "
|
| 24 |
-
"bytes":
|
| 25 |
},
|
| 26 |
"agents/agent_trajectories.jsonl": {
|
| 27 |
-
"sha256": "
|
| 28 |
-
"bytes":
|
| 29 |
},
|
| 30 |
"capture/compact.jsonl": {
|
| 31 |
-
"sha256": "
|
| 32 |
-
"bytes":
|
| 33 |
},
|
| 34 |
"capture/extended.jsonl": {
|
| 35 |
-
"sha256": "
|
| 36 |
-
"bytes":
|
| 37 |
},
|
| 38 |
"data/test.jsonl": {
|
| 39 |
-
"sha256": "
|
| 40 |
-
"bytes":
|
| 41 |
},
|
| 42 |
"data/train.jsonl": {
|
| 43 |
-
"sha256": "
|
| 44 |
-
"bytes":
|
| 45 |
},
|
| 46 |
"data/validation.jsonl": {
|
| 47 |
-
"sha256": "
|
| 48 |
-
"bytes":
|
| 49 |
},
|
| 50 |
"privacy_report.json": {
|
| 51 |
-
"sha256": "
|
| 52 |
-
"bytes":
|
| 53 |
},
|
| 54 |
"quality_report.json": {
|
| 55 |
-
"sha256": "
|
| 56 |
-
"bytes":
|
| 57 |
},
|
| 58 |
"stage2/added_2026-11-13.jsonl": {
|
| 59 |
-
"sha256": "
|
| 60 |
-
"bytes":
|
| 61 |
}
|
| 62 |
},
|
| 63 |
"version": "1.0.1",
|
| 64 |
"repo_id": "biscet/DFlash-DeepSeek-TraceMix",
|
| 65 |
"revision": "main",
|
| 66 |
-
"snapshot_revision": "
|
| 67 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"tasks": 41715,
|
| 3 |
+
"stage2_tasks": 272,
|
| 4 |
+
"combined_tasks": 41987,
|
| 5 |
"trajectories": 200,
|
| 6 |
"steps": 294,
|
| 7 |
"files": {
|
|
|
|
| 18 |
"bytes": 1689
|
| 19 |
},
|
| 20 |
"README.md": {
|
| 21 |
+
"sha256": "9249da34dcf2e82b262bcff09e9e55f69ab3ef4df37170c14aa9d987c6936806",
|
| 22 |
+
"bytes": 26535
|
| 23 |
},
|
| 24 |
"agents/agent_steps.jsonl": {
|
| 25 |
+
"sha256": "e440c7fe9b33b4d80af209faab7020fc00847f9137c18cede38ad41bbc50a990",
|
| 26 |
+
"bytes": 7026174
|
| 27 |
},
|
| 28 |
"agents/agent_trajectories.jsonl": {
|
| 29 |
+
"sha256": "68c451c2fefe8b7042a1a9b3820f571cbee48a314c02c9efa498e308b311783a",
|
| 30 |
+
"bytes": 6539665
|
| 31 |
},
|
| 32 |
"capture/compact.jsonl": {
|
| 33 |
+
"sha256": "9124c78a080ec0856d4e75ab24c53b7a3a231cb92adb4b95e8efe94159b232ae",
|
| 34 |
+
"bytes": 115627119
|
| 35 |
},
|
| 36 |
"capture/extended.jsonl": {
|
| 37 |
+
"sha256": "56dc5dcd4486ed159c0ceff1994a6de4433731c4b4673d3ec0e8a915e1b780f5",
|
| 38 |
+
"bytes": 10162310
|
| 39 |
},
|
| 40 |
"data/test.jsonl": {
|
| 41 |
+
"sha256": "051024f1fc0ff64eb1fb79d2012d7f08dc7459f77737d04ea4b27730786c7182",
|
| 42 |
+
"bytes": 12497371
|
| 43 |
},
|
| 44 |
"data/train.jsonl": {
|
| 45 |
+
"sha256": "c7b886614647c1f67a8b0db45d2b34f6af875e599e4318118af9e7c1c97a1569",
|
| 46 |
+
"bytes": 265345388
|
| 47 |
},
|
| 48 |
"data/validation.jsonl": {
|
| 49 |
+
"sha256": "5b6e6eb5fa0b068f049ffc53510ec69b297e5d3b2e07257e2b410a38dbd8a78e",
|
| 50 |
+
"bytes": 13349115
|
| 51 |
},
|
| 52 |
"privacy_report.json": {
|
| 53 |
+
"sha256": "3847ac23d0f9814931e3e81302b47b1e4ad693e71182033f25ec04f3f2798026",
|
| 54 |
+
"bytes": 979
|
| 55 |
},
|
| 56 |
"quality_report.json": {
|
| 57 |
+
"sha256": "e0d0e10d135f02af3811964088934a4019865ca6c0a3d7c6f2c0bf6149f0fc72",
|
| 58 |
+
"bytes": 8783
|
| 59 |
},
|
| 60 |
"stage2/added_2026-11-13.jsonl": {
|
| 61 |
+
"sha256": "3cd22ac15af62e5edf348e9d2a3b5138f7bc61fb45f12361566ca5f2b70312ed",
|
| 62 |
+
"bytes": 10398717
|
| 63 |
}
|
| 64 |
},
|
| 65 |
"version": "1.0.1",
|
| 66 |
"repo_id": "biscet/DFlash-DeepSeek-TraceMix",
|
| 67 |
"revision": "main",
|
| 68 |
+
"snapshot_revision": "main"
|
| 69 |
}
|
privacy_report.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"checked_at": "2026-11-13T12:
|
| 3 |
"release_version": "1.0.1",
|
| 4 |
"all_checks_passed": true,
|
| 5 |
-
"checked_jsonl_rows":
|
| 6 |
"remaining_pattern_violations": 0,
|
| 7 |
"violations_by_pattern": {},
|
| 8 |
"allowed_placeholder_matches": {
|
|
@@ -16,12 +16,7 @@
|
|
| 16 |
"rows": 272,
|
| 17 |
"changed_rows": 272,
|
| 18 |
"redactions": 1897
|
| 19 |
-
},
|
| 20 |
-
"data/train.jsonl": {
|
| 21 |
-
"rows": 37690,
|
| 22 |
-
"changed_rows": 272,
|
| 23 |
-
"redactions": 1897
|
| 24 |
}
|
| 25 |
},
|
| 26 |
-
"retained_audit": "Rows carried over from v1.0.0 were covered by the 2026-11-11 publication audit; no retained row was modified. New stage-2 rows were sanitised in this release."
|
| 27 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"checked_at": "2026-11-13T12:36:56+00:00",
|
| 3 |
"release_version": "1.0.1",
|
| 4 |
"all_checks_passed": true,
|
| 5 |
+
"checked_jsonl_rows": 84196,
|
| 6 |
"remaining_pattern_violations": 0,
|
| 7 |
"violations_by_pattern": {},
|
| 8 |
"allowed_placeholder_matches": {
|
|
|
|
| 16 |
"rows": 272,
|
| 17 |
"changed_rows": 272,
|
| 18 |
"redactions": 1897
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 19 |
}
|
| 20 |
},
|
| 21 |
+
"retained_audit": "Rows carried over from v1.0.0 were covered by the 2026-11-11 publication audit; no retained row was modified, only four literary sources were removed. New stage-2 rows were sanitised in this release."
|
| 22 |
}
|
quality_report.json
CHANGED
|
@@ -3,72 +3,69 @@
|
|
| 3 |
"target_model": "Whittle-Qwen-3.8-35B-A3B-Q5_K_M",
|
| 4 |
"draft_architecture": "DFlash2",
|
| 5 |
"seed": 641384055,
|
| 6 |
-
"target_total_tasks":
|
| 7 |
-
"completed_tasks":
|
| 8 |
"completion_ratio": 1.0,
|
| 9 |
-
"semantic_families_total":
|
| 10 |
"splits": {
|
| 11 |
"train": {
|
| 12 |
-
"tasks":
|
| 13 |
-
"semantic_families":
|
| 14 |
-
"share_tasks": 0.
|
| 15 |
-
"share_families": 0.
|
| 16 |
"languages": {
|
| 17 |
-
"en":
|
| 18 |
-
"ru":
|
| 19 |
},
|
| 20 |
"length_buckets": {
|
| 21 |
-
"compact":
|
| 22 |
-
"extended":
|
| 23 |
-
"long": 12
|
| 24 |
}
|
| 25 |
},
|
| 26 |
"validation": {
|
| 27 |
-
"tasks":
|
| 28 |
-
"semantic_families":
|
| 29 |
-
"share_tasks": 0.
|
| 30 |
-
"share_families": 0.
|
| 31 |
"languages": {
|
| 32 |
-
"en":
|
| 33 |
-
"ru":
|
| 34 |
},
|
| 35 |
"length_buckets": {
|
| 36 |
-
"compact":
|
| 37 |
-
"extended":
|
| 38 |
-
"long": 4
|
| 39 |
}
|
| 40 |
},
|
| 41 |
"test": {
|
| 42 |
-
"tasks":
|
| 43 |
-
"semantic_families":
|
| 44 |
-
"share_tasks": 0.
|
| 45 |
-
"share_families": 0.
|
| 46 |
"languages": {
|
| 47 |
-
"ru":
|
| 48 |
-
"en":
|
| 49 |
},
|
| 50 |
"length_buckets": {
|
| 51 |
-
"extended":
|
| 52 |
-
"compact": 2045
|
| 53 |
-
"long": 2
|
| 54 |
}
|
| 55 |
}
|
| 56 |
},
|
| 57 |
"language_distribution": {
|
| 58 |
-
"en":
|
| 59 |
-
"ru":
|
| 60 |
},
|
| 61 |
-
"native_russian_count":
|
| 62 |
"translated_russian_count": 0,
|
| 63 |
"native_russian_ratio": 1.0,
|
| 64 |
"source_kind_distribution": {
|
| 65 |
-
"local_project":
|
| 66 |
"public_domain_text": 21313,
|
| 67 |
"open_dataset": 1262,
|
| 68 |
"self_created": 58
|
| 69 |
},
|
| 70 |
"domain_distribution": {
|
| 71 |
-
"frontend_ui":
|
| 72 |
"technical_writing": 12989,
|
| 73 |
"general_knowledge": 4484,
|
| 74 |
"systems_programming": 19,
|
|
@@ -82,12 +79,12 @@
|
|
| 82 |
"security_reliability": 9
|
| 83 |
},
|
| 84 |
"programming_language_distribution": {
|
| 85 |
-
"TypeScript":
|
| 86 |
-
"general":
|
| 87 |
"C/C++": 23,
|
| 88 |
"React": 2783,
|
| 89 |
-
"JavaScript":
|
| 90 |
-
"Vue":
|
| 91 |
"Go": 1044,
|
| 92 |
"HTML/CSS": 836,
|
| 93 |
"SQL": 174,
|
|
@@ -97,91 +94,112 @@
|
|
| 97 |
"Server/Infrastructure": 28
|
| 98 |
},
|
| 99 |
"task_type_distribution": {
|
| 100 |
-
"api_and_database":
|
| 101 |
"testing": 2592,
|
| 102 |
"text_processing": 12989,
|
| 103 |
"general_qa": 4486,
|
| 104 |
-
"code_explanation":
|
| 105 |
-
"refactoring":
|
| 106 |
-
"bug_finding":
|
| 107 |
-
"patch_preparation":
|
| 108 |
"type_fixing": 2458,
|
| 109 |
"logic_reasoning": 4251,
|
| 110 |
-
"async_and_concurrency":
|
| 111 |
"function_implementation": 366,
|
| 112 |
-
"agent_next_action":
|
| 113 |
"math_reasoning": 196,
|
| 114 |
"multi_file_analysis": 14,
|
| 115 |
"configuration": 26,
|
| 116 |
"log_analysis": 6
|
| 117 |
},
|
| 118 |
"difficulty_distribution": {
|
| 119 |
-
"hard":
|
| 120 |
-
"medium":
|
| 121 |
-
"easy":
|
| 122 |
},
|
| 123 |
"context_size_distribution": {
|
| 124 |
-
"long":
|
| 125 |
-
"short":
|
| 126 |
-
"medium":
|
| 127 |
"multi_file": 14
|
| 128 |
},
|
| 129 |
"response_format_distribution": {
|
| 130 |
-
"code_and_explanation":
|
| 131 |
"test_suite": 1373,
|
| 132 |
"concise_answer": 12464,
|
| 133 |
-
"structured_analysis":
|
| 134 |
-
"unified_diff_patch":
|
| 135 |
-
"root_cause_and_fix":
|
| 136 |
"step_by_step_solution": 2414,
|
| 137 |
-
"tool_call":
|
| 138 |
},
|
| 139 |
-
"software_development_tasks":
|
| 140 |
-
"software_development_share": 0.
|
| 141 |
"text_and_reasoning_tasks": 21922,
|
| 142 |
-
"text_and_reasoning_share": 0.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 143 |
"length_buckets": {
|
| 144 |
-
"compact":
|
| 145 |
-
"extended":
|
| 146 |
},
|
| 147 |
"task_length_buckets": {
|
| 148 |
-
"compact":
|
| 149 |
-
"extended":
|
| 150 |
-
"long": 18
|
| 151 |
},
|
| 152 |
"token_length_approximate": false,
|
| 153 |
"length_statistics": {
|
| 154 |
"all": {
|
| 155 |
-
"count":
|
| 156 |
"min_tokens": 52,
|
| 157 |
-
"max_tokens":
|
| 158 |
-
"mean_tokens":
|
| 159 |
-
"median_tokens":
|
| 160 |
-
"p95_tokens":
|
| 161 |
},
|
| 162 |
"compact": {
|
| 163 |
-
"count":
|
| 164 |
"min_tokens": 52,
|
| 165 |
"max_tokens": 1536,
|
| 166 |
-
"mean_tokens": 547.
|
| 167 |
-
"median_tokens":
|
| 168 |
-
"p95_tokens":
|
| 169 |
},
|
| 170 |
"extended": {
|
| 171 |
-
"count":
|
| 172 |
"min_tokens": 1538,
|
| 173 |
-
"max_tokens":
|
| 174 |
-
"mean_tokens":
|
| 175 |
-
"median_tokens":
|
| 176 |
-
"p95_tokens":
|
| 177 |
},
|
| 178 |
"long": {
|
| 179 |
-
"count":
|
| 180 |
-
"min_tokens": 8197,
|
| 181 |
-
"max_tokens": 12967,
|
| 182 |
-
"mean_tokens": 9423.22,
|
| 183 |
-
"median_tokens": 8963.0,
|
| 184 |
-
"p95_tokens": 12918
|
| 185 |
}
|
| 186 |
},
|
| 187 |
"duplicates_and_rejections": {
|
|
@@ -190,24 +208,20 @@
|
|
| 190 |
"rejected_total": 0,
|
| 191 |
"rejection_reasons": {},
|
| 192 |
"same_family_variants_allowed": true,
|
| 193 |
-
"unique_user_texts":
|
| 194 |
},
|
| 195 |
"content_diversity_metrics": {
|
| 196 |
-
"unique_source_groups":
|
| 197 |
-
"unique_user_task_hashes":
|
| 198 |
-
"unique_instruction_prefixes":
|
| 199 |
-
"type_token_ratio": 0.
|
| 200 |
"max_family_size": 10,
|
| 201 |
-
"mean_tasks_per_family": 4.
|
| 202 |
-
"families_with_multiple_tasks":
|
| 203 |
},
|
| 204 |
"privacy_and_decontamination": {
|
| 205 |
-
"pre_tokenization_redactions":
|
| 206 |
"hard_violations": {},
|
| 207 |
-
"cross_split": {
|
| 208 |
-
"heldout_exclusive_13grams": 217459,
|
| 209 |
-
"contaminated_new_train_rows": 0
|
| 210 |
-
},
|
| 211 |
"secrets_or_pii_found": false
|
| 212 |
},
|
| 213 |
"release_changes": {
|
|
@@ -220,9 +234,7 @@
|
|
| 220 |
"removed_text_tasks": 8285,
|
| 221 |
"added_stage2_tasks": 272,
|
| 222 |
"stage2_split_counts": {
|
| 223 |
-
"train":
|
| 224 |
-
"validation": 42,
|
| 225 |
-
"test": 40
|
| 226 |
},
|
| 227 |
"stage2_tag": "stage2-2026-11-13",
|
| 228 |
"stage2_source_files": "stage2/added_2026-11-13.jsonl"
|
|
@@ -231,22 +243,22 @@
|
|
| 231 |
{
|
| 232 |
"check_name": "schema_valid",
|
| 233 |
"passed": true,
|
| 234 |
-
"details": "records=
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"check_name": "split_family_disjoint",
|
| 238 |
"passed": true,
|
| 239 |
-
"details": "families=
|
| 240 |
},
|
| 241 |
{
|
| 242 |
"check_name": "split_ratio_90_5_5",
|
| 243 |
"passed": true,
|
| 244 |
-
"details": "{\"test\":
|
| 245 |
},
|
| 246 |
{
|
| 247 |
"check_name": "language_balance_45_55",
|
| 248 |
"passed": true,
|
| 249 |
-
"details": "{\"en\":
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"check_name": "native_russian_share",
|
|
@@ -256,12 +268,12 @@
|
|
| 256 |
{
|
| 257 |
"check_name": "software_development_share",
|
| 258 |
"passed": true,
|
| 259 |
-
"details": "share=0.
|
| 260 |
},
|
| 261 |
{
|
| 262 |
"check_name": "text_and_reasoning_share",
|
| 263 |
"passed": true,
|
| 264 |
-
"details": "share=0.
|
| 265 |
},
|
| 266 |
{
|
| 267 |
"check_name": "no_assistant_completion",
|
|
@@ -301,7 +313,7 @@
|
|
| 301 |
{
|
| 302 |
"check_name": "length_buckets_consistent",
|
| 303 |
"passed": true,
|
| 304 |
-
"details": "{\"tasks\": {\"compact\":
|
| 305 |
},
|
| 306 |
{
|
| 307 |
"check_name": "capture_covers_bounded_prompts",
|
|
@@ -311,7 +323,7 @@
|
|
| 311 |
{
|
| 312 |
"check_name": "prompt_length_bounds",
|
| 313 |
"passed": true,
|
| 314 |
-
"details": "min=52, max=
|
| 315 |
},
|
| 316 |
{
|
| 317 |
"check_name": "sources_present",
|
|
@@ -329,9 +341,9 @@
|
|
| 329 |
"details": "converted=272, path_and_project_redactions=1897"
|
| 330 |
},
|
| 331 |
{
|
| 332 |
-
"check_name": "
|
| 333 |
"passed": true,
|
| 334 |
-
"details": "
|
| 335 |
}
|
| 336 |
],
|
| 337 |
"all_checks_passed": true
|
|
|
|
| 3 |
"target_model": "Whittle-Qwen-3.8-35B-A3B-Q5_K_M",
|
| 4 |
"draft_architecture": "DFlash2",
|
| 5 |
"seed": 641384055,
|
| 6 |
+
"target_total_tasks": 41715,
|
| 7 |
+
"completed_tasks": 41715,
|
| 8 |
"completion_ratio": 1.0,
|
| 9 |
+
"semantic_families_total": 10076,
|
| 10 |
"splits": {
|
| 11 |
"train": {
|
| 12 |
+
"tasks": 37500,
|
| 13 |
+
"semantic_families": 8398,
|
| 14 |
+
"share_tasks": 0.898957,
|
| 15 |
+
"share_families": 0.833466,
|
| 16 |
"languages": {
|
| 17 |
+
"en": 19059,
|
| 18 |
+
"ru": 18441
|
| 19 |
},
|
| 20 |
"length_buckets": {
|
| 21 |
+
"compact": 36234,
|
| 22 |
+
"extended": 1266
|
|
|
|
| 23 |
}
|
| 24 |
},
|
| 25 |
"validation": {
|
| 26 |
+
"tasks": 2110,
|
| 27 |
+
"semantic_families": 841,
|
| 28 |
+
"share_tasks": 0.050581,
|
| 29 |
+
"share_families": 0.083466,
|
| 30 |
"languages": {
|
| 31 |
+
"en": 1013,
|
| 32 |
+
"ru": 1097
|
| 33 |
},
|
| 34 |
"length_buckets": {
|
| 35 |
+
"compact": 2016,
|
| 36 |
+
"extended": 94
|
|
|
|
| 37 |
}
|
| 38 |
},
|
| 39 |
"test": {
|
| 40 |
+
"tasks": 2105,
|
| 41 |
+
"semantic_families": 837,
|
| 42 |
+
"share_tasks": 0.050461,
|
| 43 |
+
"share_families": 0.083069,
|
| 44 |
"languages": {
|
| 45 |
+
"ru": 1008,
|
| 46 |
+
"en": 1097
|
| 47 |
},
|
| 48 |
"length_buckets": {
|
| 49 |
+
"extended": 60,
|
| 50 |
+
"compact": 2045
|
|
|
|
| 51 |
}
|
| 52 |
}
|
| 53 |
},
|
| 54 |
"language_distribution": {
|
| 55 |
+
"en": 21169,
|
| 56 |
+
"ru": 20546
|
| 57 |
},
|
| 58 |
+
"native_russian_count": 20546,
|
| 59 |
"translated_russian_count": 0,
|
| 60 |
"native_russian_ratio": 1.0,
|
| 61 |
"source_kind_distribution": {
|
| 62 |
+
"local_project": 19082,
|
| 63 |
"public_domain_text": 21313,
|
| 64 |
"open_dataset": 1262,
|
| 65 |
"self_created": 58
|
| 66 |
},
|
| 67 |
"domain_distribution": {
|
| 68 |
+
"frontend_ui": 17860,
|
| 69 |
"technical_writing": 12989,
|
| 70 |
"general_knowledge": 4484,
|
| 71 |
"systems_programming": 19,
|
|
|
|
| 79 |
"security_reliability": 9
|
| 80 |
},
|
| 81 |
"programming_language_distribution": {
|
| 82 |
+
"TypeScript": 10642,
|
| 83 |
+
"general": 22157,
|
| 84 |
"C/C++": 23,
|
| 85 |
"React": 2783,
|
| 86 |
+
"JavaScript": 2077,
|
| 87 |
+
"Vue": 1797,
|
| 88 |
"Go": 1044,
|
| 89 |
"HTML/CSS": 836,
|
| 90 |
"SQL": 174,
|
|
|
|
| 94 |
"Server/Infrastructure": 28
|
| 95 |
},
|
| 96 |
"task_type_distribution": {
|
| 97 |
+
"api_and_database": 1859,
|
| 98 |
"testing": 2592,
|
| 99 |
"text_processing": 12989,
|
| 100 |
"general_qa": 4486,
|
| 101 |
+
"code_explanation": 3085,
|
| 102 |
+
"refactoring": 3071,
|
| 103 |
+
"bug_finding": 2727,
|
| 104 |
+
"patch_preparation": 3108,
|
| 105 |
"type_fixing": 2458,
|
| 106 |
"logic_reasoning": 4251,
|
| 107 |
+
"async_and_concurrency": 460,
|
| 108 |
"function_implementation": 366,
|
| 109 |
+
"agent_next_action": 21,
|
| 110 |
"math_reasoning": 196,
|
| 111 |
"multi_file_analysis": 14,
|
| 112 |
"configuration": 26,
|
| 113 |
"log_analysis": 6
|
| 114 |
},
|
| 115 |
"difficulty_distribution": {
|
| 116 |
+
"hard": 5181,
|
| 117 |
+
"medium": 19296,
|
| 118 |
+
"easy": 17238
|
| 119 |
},
|
| 120 |
"context_size_distribution": {
|
| 121 |
+
"long": 7538,
|
| 122 |
+
"short": 16111,
|
| 123 |
+
"medium": 18052,
|
| 124 |
"multi_file": 14
|
| 125 |
},
|
| 126 |
"response_format_distribution": {
|
| 127 |
+
"code_and_explanation": 6221,
|
| 128 |
"test_suite": 1373,
|
| 129 |
"concise_answer": 12464,
|
| 130 |
+
"structured_analysis": 11272,
|
| 131 |
+
"unified_diff_patch": 5512,
|
| 132 |
+
"root_cause_and_fix": 2438,
|
| 133 |
"step_by_step_solution": 2414,
|
| 134 |
+
"tool_call": 21
|
| 135 |
},
|
| 136 |
+
"software_development_tasks": 19793,
|
| 137 |
+
"software_development_share": 0.474482,
|
| 138 |
"text_and_reasoning_tasks": 21922,
|
| 139 |
+
"text_and_reasoning_share": 0.525518,
|
| 140 |
+
"stage2_tasks": 272,
|
| 141 |
+
"combined_tasks": 41987,
|
| 142 |
+
"stage2_language_distribution": {
|
| 143 |
+
"ru": 139,
|
| 144 |
+
"en": 133
|
| 145 |
+
},
|
| 146 |
+
"stage2_task_type_distribution": {
|
| 147 |
+
"patch_preparation": 22,
|
| 148 |
+
"api_and_database": 28,
|
| 149 |
+
"refactoring": 10,
|
| 150 |
+
"async_and_concurrency": 7,
|
| 151 |
+
"bug_finding": 41,
|
| 152 |
+
"code_explanation": 152,
|
| 153 |
+
"agent_next_action": 12
|
| 154 |
+
},
|
| 155 |
+
"agent_splits": {
|
| 156 |
+
"agent_trajectories": {
|
| 157 |
+
"validation": 10,
|
| 158 |
+
"train": 180,
|
| 159 |
+
"test": 10
|
| 160 |
+
},
|
| 161 |
+
"agent_steps": {
|
| 162 |
+
"validation": 14,
|
| 163 |
+
"train": 268,
|
| 164 |
+
"test": 12
|
| 165 |
+
}
|
| 166 |
+
},
|
| 167 |
"length_buckets": {
|
| 168 |
+
"compact": 40295,
|
| 169 |
+
"extended": 1420
|
| 170 |
},
|
| 171 |
"task_length_buckets": {
|
| 172 |
+
"compact": 40295,
|
| 173 |
+
"extended": 1420
|
|
|
|
| 174 |
},
|
| 175 |
"token_length_approximate": false,
|
| 176 |
"length_statistics": {
|
| 177 |
"all": {
|
| 178 |
+
"count": 41715,
|
| 179 |
"min_tokens": 52,
|
| 180 |
+
"max_tokens": 5030,
|
| 181 |
+
"mean_tokens": 589.07,
|
| 182 |
+
"median_tokens": 453,
|
| 183 |
+
"p95_tokens": 1423
|
| 184 |
},
|
| 185 |
"compact": {
|
| 186 |
+
"count": 40295,
|
| 187 |
"min_tokens": 52,
|
| 188 |
"max_tokens": 1536,
|
| 189 |
+
"mean_tokens": 547.19,
|
| 190 |
+
"median_tokens": 446,
|
| 191 |
+
"p95_tokens": 1231
|
| 192 |
},
|
| 193 |
"extended": {
|
| 194 |
+
"count": 1420,
|
| 195 |
"min_tokens": 1538,
|
| 196 |
+
"max_tokens": 5030,
|
| 197 |
+
"mean_tokens": 1777.45,
|
| 198 |
+
"median_tokens": 1713.5,
|
| 199 |
+
"p95_tokens": 2183
|
| 200 |
},
|
| 201 |
"long": {
|
| 202 |
+
"count": 0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 203 |
}
|
| 204 |
},
|
| 205 |
"duplicates_and_rejections": {
|
|
|
|
| 208 |
"rejected_total": 0,
|
| 209 |
"rejection_reasons": {},
|
| 210 |
"same_family_variants_allowed": true,
|
| 211 |
+
"unique_user_texts": 41715
|
| 212 |
},
|
| 213 |
"content_diversity_metrics": {
|
| 214 |
+
"unique_source_groups": 10076,
|
| 215 |
+
"unique_user_task_hashes": 41715,
|
| 216 |
+
"unique_instruction_prefixes": 1007,
|
| 217 |
+
"type_token_ratio": 0.014172,
|
| 218 |
"max_family_size": 10,
|
| 219 |
+
"mean_tasks_per_family": 4.14,
|
| 220 |
+
"families_with_multiple_tasks": 8691
|
| 221 |
},
|
| 222 |
"privacy_and_decontamination": {
|
| 223 |
+
"pre_tokenization_redactions": 10349,
|
| 224 |
"hard_violations": {},
|
|
|
|
|
|
|
|
|
|
|
|
|
| 225 |
"secrets_or_pii_found": false
|
| 226 |
},
|
| 227 |
"release_changes": {
|
|
|
|
| 234 |
"removed_text_tasks": 8285,
|
| 235 |
"added_stage2_tasks": 272,
|
| 236 |
"stage2_split_counts": {
|
| 237 |
+
"train": 272
|
|
|
|
|
|
|
| 238 |
},
|
| 239 |
"stage2_tag": "stage2-2026-11-13",
|
| 240 |
"stage2_source_files": "stage2/added_2026-11-13.jsonl"
|
|
|
|
| 243 |
{
|
| 244 |
"check_name": "schema_valid",
|
| 245 |
"passed": true,
|
| 246 |
+
"details": "records=41715, unique_ids=True"
|
| 247 |
},
|
| 248 |
{
|
| 249 |
"check_name": "split_family_disjoint",
|
| 250 |
"passed": true,
|
| 251 |
+
"details": "families=10076, multi_split_families=0"
|
| 252 |
},
|
| 253 |
{
|
| 254 |
"check_name": "split_ratio_90_5_5",
|
| 255 |
"passed": true,
|
| 256 |
+
"details": "{\"test\": 2105, \"train\": 37500, \"validation\": 2110}"
|
| 257 |
},
|
| 258 |
{
|
| 259 |
"check_name": "language_balance_45_55",
|
| 260 |
"passed": true,
|
| 261 |
+
"details": "{\"en\": 21169, \"ru\": 20546}"
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"check_name": "native_russian_share",
|
|
|
|
| 268 |
{
|
| 269 |
"check_name": "software_development_share",
|
| 270 |
"passed": true,
|
| 271 |
+
"details": "share=0.474"
|
| 272 |
},
|
| 273 |
{
|
| 274 |
"check_name": "text_and_reasoning_share",
|
| 275 |
"passed": true,
|
| 276 |
+
"details": "share=0.526"
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"check_name": "no_assistant_completion",
|
|
|
|
| 313 |
{
|
| 314 |
"check_name": "length_buckets_consistent",
|
| 315 |
"passed": true,
|
| 316 |
+
"details": "{\"tasks\": {\"compact\": 40295, \"extended\": 1420}, \"capture\": {\"compact\": 40295, \"extended\": 1420}}"
|
| 317 |
},
|
| 318 |
{
|
| 319 |
"check_name": "capture_covers_bounded_prompts",
|
|
|
|
| 323 |
{
|
| 324 |
"check_name": "prompt_length_bounds",
|
| 325 |
"passed": true,
|
| 326 |
+
"details": "min=52, max=5030"
|
| 327 |
},
|
| 328 |
{
|
| 329 |
"check_name": "sources_present",
|
|
|
|
| 341 |
"details": "converted=272, path_and_project_redactions=1897"
|
| 342 |
},
|
| 343 |
{
|
| 344 |
+
"check_name": "stage2_separate_from_tasks",
|
| 345 |
"passed": true,
|
| 346 |
+
"details": "stage2_records=272, task_records=41715"
|
| 347 |
}
|
| 348 |
],
|
| 349 |
"all_checks_passed": true
|
stage2/added_2026-11-13.jsonl
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|