INDEX

Explanations

words related to visual appearance or perception

oai_token-act-pair · gpt-3.5-turbo

instances of the verb "appear" and its variations indicating visibility or existence

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

mys

-0.71

ses

-0.65

osure

-0.63

ests

-0.61

zsche

-0.60

oller

-0.59

dp

-0.58

milo

-0.58

shaw

-0.58

mong

-0.57

POSITIVE LOGITS

 poised

0.92

to

0.82

 destined

0.79

 prominently

0.75

 innocuous

0.74

 unstoppable

0.70

 unchanged

0.70

 dormant

0.69

 spor

0.68

 victorious

0.68

Activations Density 0.052%

words related to visual appearance or perception

instances of the verb "appear" and its variations indicating visibility or existence

No Comments

No Known Activations