wehub-resource-sync
17 lines
1.3 KiB
Plaintext
17 lines
1.3 KiB
Plaintext
flowchart LR
|
|
A1([Pretrain: The Pile]):::data --> A2[tiktoken r50k_base<br/>stream + encode]:::proc --> A3[(pile_train.h5<br/>flat int32 tokens)]:::store
|
|
B1([SFT: Alpaca · Dolly · GSM8K]):::data --> B2[chat template<br/>+ mask the prompt]:::proc --> B3[(sft_packed.h5<br/>tokens + loss_mask)]:::store
|
|
C1([Prefs: HH-RLHF · UltraFeedback]):::data --> C2[split prompt /<br/>chosen / rejected]:::proc --> C3[(preferences.jsonl)]:::store
|
|
D1([RL: GSM8K · arithmetic]):::data --> D2[keep the gold answer]:::proc --> D3[(rl_prompts.jsonl)]:::store
|
|
|
|
classDef data fill:#cdeccd,stroke:#2e7d32,stroke-width:2px,color:#143d1a;
|
|
classDef store fill:#cfeee9,stroke:#1f9e8f,stroke-width:2px,color:#0c3a34;
|
|
classDef proc fill:#cfe3fb,stroke:#1565c0,stroke-width:2px,color:#0d2c52;
|
|
classDef model fill:#ffe8a3,stroke:#d48806,stroke-width:2px,color:#5a3d00;
|
|
classDef rl fill:#ffd9b3,stroke:#e67e22,stroke-width:2px,color:#6b3500;
|
|
classDef loss fill:#ffd4d4,stroke:#c0392b,stroke-width:2px,color:#5c1212;
|
|
classDef ckpt fill:#ececec,stroke:#666666,stroke-width:2px,color:#222222;
|
|
classDef eval fill:#e7d6fb,stroke:#8e44ad,stroke-width:2px,color:#3d1a5a;
|
|
classDef out fill:#e7d6fb,stroke:#7b3fbf,stroke-width:2px,color:#3d1a5a;
|
|
classDef op fill:#ffffff,stroke:#888888,stroke-width:2px,color:#333333;
|