INDEX

Explanations

numerical codes or references to specific locations

oai_token-act-pair · gpt-3.5-turbo

numerical values or sequences, particularly associated with specific categories or data points

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

oru

-0.76

gart

-0.75

giving

-0.75

hatt

-0.74

gewater

-0.73

estine

-0.72

bara

-0.72

gone

-0.70

swick

-0.68

lace

-0.68

POSITIVE LOGITS

 ILCS

1.01

mph

0.84

dB

0.79

0.76

MPH

0.73

 entrants

0.72

estinal

0.72

Activations Density 0.021%

numerical codes or references to specific locations

numerical values or sequences, particularly associated with specific categories or data points

No Comments

No Known Activations