metadata

license: apache-2.0
tags:
  - StepLaw
  - causal-lm
language:
  - en
library_name: transformers
pipeline_tag: text-generation
model-index:
  - name: >-
      step2v2_0618_h1024_ffnh9552_numh16_numl8_lr4.883e-04_bs64_ti610351_mlr1.00e-05
    results: []

Wandb Model Name: step2v2_0618_h1024_ffnh9552_numh16_numl8_lr4.883e-04_bs64_ti610351_mlr1.00e-05

This model is part of the StepLaw-N_268M-D_79.0B collection.

Model Specifications

Architecture

Hidden size (H): 1024
Feed-forward network size (FFN): 9552
Attention heads: 16
Layers: 8
Parameter count: 268M

Training Parameters

Learning rate (lr): 4.883e-04
Batch size (bs): 131072
Training iterations: 610351
Training tokens (D): 80.0B

Model Description

StepLaw models are trained with various hyperparameter settings to enable research on scaling laws and hyperparameter optimization. This specific model was trained with learning rate 4.883e-04 and batch size 131072 for 610351 iterations, using a total of 80.0B training tokens.

Usage Example

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "StepLaw/StepLaw-N_268M-D_79.0B-LR4.883e-04-BS131072"
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True, use_fast=False)
model = AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=True)

# Generate text
inputs = tokenizer("A long time ago in a galaxy far, far away", return_tensors="pt")
outputs = model.generate(**inputs, max_length=100)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))