Hugging Face's logo

Datasets:
TokenBender
/
glm47-pie-cpp-posttraining-data

Tasks:
Text Generation
Modalities:
Text
Formats:
json
Languages:
English
Size:
10K - 100K
Tags:
code
code-optimization
cpp
sft
grpo
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
TokenBender commited on
Commit
5903856
·
verified ·
1 Parent(s): 5bb3330

data: include self-contained PIE task bundle

Browse files
Files changed (3) hide show
  1. README.md +4 -2
  2. SHA256SUMS +2 -1
  3. tasks.tar.gz +3 -0
README.md CHANGED
@@ -39,10 +39,13 @@ post-training runs.
39
  | `eval/validation.jsonl` | 1,259 | Full held-out evaluation |
40
  | `eval/validation_mini126.jsonl` | 126 | Fast evaluation |
41
  | `eval/validation_mini4.jsonl` | 4 | Smoke evaluation |
 
42
 
43
  Each row carries a stable task ID, problem ID, split, prompt, label, and task
44
  metadata. SFT rows additionally contain the user and assistant messages used by
45
- the trainer.
 
 
46
 
47
  ## Preparation
48
 
@@ -62,4 +65,3 @@ applicable upstream dataset terms.
62
 
63
  Training code:
64
  [TokenBender/browser-is-all-you-need](https://github.com/tokenbender/browser-is-all-you-need/tree/client/glm47-h100-posttraining)
65
-
 
39
  | `eval/validation.jsonl` | 1,259 | Full held-out evaluation |
40
  | `eval/validation_mini126.jsonl` | 126 | Fast evaluation |
41
  | `eval/validation_mini4.jsonl` | 4 | Smoke evaluation |
42
+ | `tasks.tar.gz` | 9,611 task JSONs | Reward scoring and evaluation harness inputs |
43
 
44
  Each row carries a stable task ID, problem ID, split, prompt, label, and task
45
  metadata. SFT rows additionally contain the user and assistant messages used by
46
+ the trainer. The task archive contains every `tasks/...` path referenced by the
47
+ training and evaluation rows. The canonical repository downloader verifies and
48
+ extracts it into `data/tasks`.
49
 
50
  ## Preparation
51
 
 
65
 
66
  Training code:
67
  [TokenBender/browser-is-all-you-need](https://github.com/tokenbender/browser-is-all-you-need/tree/client/glm47-h100-posttraining)
 
SHA256SUMS CHANGED
@@ -1,7 +1,8 @@
1
- 62587948eb1e5e56a8ad40e120f1ff2158632d58af954f3a37c6bd03f822d562 ./README.md
2
  9297d35ed42a807239bcea812931045673475e4389220ea0b732115788dade6e ./eval/validation.jsonl
3
  e1efd4bc975310dd34946c4ec9705187f82d67e7ab30d41643f063b48b372929 ./eval/validation_mini126.jsonl
4
  83ea467298740f35f49242c909c9f8d636c7f23ac69d62dac8e3bc106a69d33f ./eval/validation_mini4.jsonl
5
  216e961127630ef687a2908b8e822d17f39dc248fcb1ab912840e15598ab1101 ./grpo/train.jsonl
6
  9660523f3044ee30b73a01d68f516a1b08a3a404abec4fa653cc43e0e1ab0cd0 ./manifest.json
7
  047007ba4f575ae694d29ef9cb8294b31fc1e7263e6c21a85be58fb81ba7afa6 ./sft/train.jsonl
 
 
1
+ f863bb31ffb511006821c23c5a406dc05807ef0c54e7765fcdcd75183ac94da3 ./README.md
2
  9297d35ed42a807239bcea812931045673475e4389220ea0b732115788dade6e ./eval/validation.jsonl
3
  e1efd4bc975310dd34946c4ec9705187f82d67e7ab30d41643f063b48b372929 ./eval/validation_mini126.jsonl
4
  83ea467298740f35f49242c909c9f8d636c7f23ac69d62dac8e3bc106a69d33f ./eval/validation_mini4.jsonl
5
  216e961127630ef687a2908b8e822d17f39dc248fcb1ab912840e15598ab1101 ./grpo/train.jsonl
6
  9660523f3044ee30b73a01d68f516a1b08a3a404abec4fa653cc43e0e1ab0cd0 ./manifest.json
7
  047007ba4f575ae694d29ef9cb8294b31fc1e7263e6c21a85be58fb81ba7afa6 ./sft/train.jsonl
8
+ 0073ece0f58d6e599fe8a72348fca8b62921c3b6fb53bba12192a712c52e0f59 ./tasks.tar.gz
tasks.tar.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0073ece0f58d6e599fe8a72348fca8b62921c3b6fb53bba12192a712c52e0f59
3
+ size 8337216