Datasetpublic

laion/CoderForge-Preview-v3-316

laion/CoderForge-Preview-v3-316 Row-subset of the pre-tokenized trajectories in togethercomputer/CoderForge-Preview (trajectories-tokenized_qwencoder subset). Size: 316 rows (source: 155,144 across 4 slugs). Format: native pre-tokenized data for Qwen3 (tokenizer shared with Qwen2.5-Coder / Qwen3-Coder / Qwen3-8B). Per row columns: input_ids: list[int32] attention_mask: list[int8] (all 1s; added by this subsetter so axolotl's auto-detection of pre-tokenized datasets triggers —… See the full description on the dataset page: https://huggingface.co/datasets/laion/CoderForge-Preview-v3-316.

sourceHugging Faceapache-2.0updated 5mo agoView on Hugging Face

0likes14downloads

rootREADME.md

README.md45 linesDownload Raw Back to root

1---2license: apache-2.03task_categories:4  - text-generation5tags:6  - sft7  - agent8  - swe-bench9  - axolotl10  - pretokenized11---12 13# laion/CoderForge-Preview-v3-31614 15Row-subset of the pre-tokenized trajectories in16[togethercomputer/CoderForge-Preview](https://huggingface.co/datasets/togethercomputer/CoderForge-Preview)17(`trajectories-tokenized_qwencoder` subset).18 19**Size**: 316 rows (source: 155,144 across 4 slugs).20 21**Format**: native pre-tokenized data for Qwen3 (tokenizer shared with Qwen2.5-Coder / Qwen3-Coder / Qwen3-8B).22Per row columns:23- `input_ids: list[int32]`24- `attention_mask: list[int8]` (all 1s; added by this subsetter so axolotl's25  auto-detection of pre-tokenized datasets triggers — upstream only had26  input_ids + labels)27- `labels: list[int64]` (with `-100` masks already applied)28- `chat_template_applied: str` (decoded render for debugging)29- `trajectory_id: str`, `reward: float64`30- `source: str` (always `"togethercomputer/CoderForge-Preview/trajectories-tokenized_qwencoder"`)31 32Sampled deterministically (seed=42) from a concatenation of all 4 source slugs33(R2E_Gym, SWE_Rebench, SWE_Smith, filtered_reward1). Row subsets are nested.34 35## Usage (axolotl)36 37```yaml38datasets:39  - path: laion/CoderForge-Preview-v3-31640chat_template: chatml41sequence_len: 32768   # sequences in the upstream data can exceed 80k tokens; axolotl truncates42```43 44Axolotl detects the pre-tokenized columns and skips the chat_template renderer.45