File size: 731 Bytes
3b2d368
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
# TinyGSM Dataset Configuration
# Math reasoning dataset for language model training

defaults:
  - _base
  - _self_

# Dataset Identity
dataset_name: tinygsm_no_slides
init_fn: lmr.data.tinygsm_no_slides.initialize_dataset

# Token Limits
max_tokens_train: None       # 50M tokens for training
max_tokens_validation: None  # 500K tokens for validation
max_tokens_test: None        # 500K tokens for test

tokens_buffer: 10k           # Additional tokens buffer to prevent inconsistencies

# TinyGSM has single component (math problems)
sampling_type: single        # Not proportional, single component
component_name: tinygsm_no_slides

# HuggingFace dataset path
hf_dataset_path: "TinyGSM/TinyGSM"  # or your TinyGSM dataset path