INDEX

Explanations

words related to directional movement or conflicting actions

oai_token-act-pair · gpt-3.5-turbo

references to movement or exchange, particularly in a back-and-forth context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ĨĴ

-0.84

ially

-0.76

ãĥĨ

-0.73

payer

-0.66

Catalog

-0.66

 payable

-0.65

MO

-0.65

ccording

-0.64

atile

-0.63

 embodied

-0.63

POSITIVE LOGITS

zig

0.82

 Tide

0.81

 shores

0.79

 throats

0.77

 whence

0.75

 corners

0.69

 Gret

0.69

 shore

0.68

 Fury

0.65

forth

0.65

Activations Density 0.463%

words related to directional movement or conflicting actions

references to movement or exchange, particularly in a back-and-forth context

No Comments

No Known Activations