INDEX

Explanations

words related to rankings or evaluations

oai_token-act-pair · gpt-3.5-turbo

instances of the word "est" and its variations, which may indicate evaluations or descriptions in a comparative context

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

gb

-0.70

igel

-0.68

Guide

-0.68

intestinal

-0.67

jee

-0.66

clerosis

-0.66

urdue

-0.65

Berry

-0.65

zzle

-0.65

Interstitial

-0.65

POSITIVE LOGITS

est

0.90

hetics

0.89

rus

0.82

hetic

0.78

inas

0.74

hetically

0.72

rant

0.71

uary

0.69

ot

0.69

 Karin

0.69

Activations Density 0.017%

words related to rankings or evaluations

instances of the word "est" and its variations, which may indicate evaluations or descriptions in a comparative context

No Known Activations