INDEX

Explanations

mentions of particular phrases or terms in different languages, potentially related to specific subjects or locations

oai_token-act-pair · gpt-3.5-turbo

references to categories or attributes of objects, particularly in a descriptive context

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

icz

-0.93

estine

-0.91

rency

-0.80

itzer

-0.78

erate

-0.78

atform

-0.78

ileaks

-0.75

Walker

-0.72

usher

-0.72

strate

-0.71

POSITIVE LOGITS

 redes

0.80

 spirits

0.78

 breaths

0.76

 possessions

0.76

ources

0.76

 illusions

0.75

 tiers

0.74

 pens

0.74

 edits

0.73

 alle

0.73

Activations Density 0.045%

mentions of particular phrases or terms in different languages, potentially related to specific subjects or locations

references to categories or attributes of objects, particularly in a descriptive context

No Comments

No Known Activations