INDEX

Explanations

phrases related to performing follow-up actions or investigations

oai_token-act-pair · gpt-3.5-turbo

references to follow-up actions or events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Scotia

-0.75

 nerv

-0.67

ãĥĨãĤ£

-0.65

Fey

-0.65

 sidx

-0.64

inese

-0.63

 ILCS

-0.60

 wounding

-0.58

 Unic

-0.58

pite

-0.56

POSITIVE LOGITS

ings

1.04

ership

0.99

ences

0.90

ingly

0.86

bys

0.82

ance

0.81

overs

0.81

ers

0.80

ents

0.80

ictions

0.78

Activations Density 0.023%

phrases related to performing follow-up actions or investigations

references to follow-up actions or events

No Comments

No Known Activations