INDEX

Explanations

phrases indicating negative or undesirable situations

oai_token-act-pair · gpt-3.5-turbo

statements that emphasize negation or denial

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

aneers

-0.88

ãĤ¨ãĥ«

-0.86

ngth

-0.84

anamo

-0.75

seys

-0.74

ãĥ¼ãĥ³

-0.74

alties

-0.74

opez

-0.73

ãĤ¦ãĤ¹

-0.73

dies

-0.72

POSITIVE LOGITS

 blat

0.77

 spac

0.71

 continuation

0.71

 generational

0.70

 happening

0.70

 whistlebl

0.69

 STEM

0.69

 blatant

0.66

 chance

0.65

 textbook

0.64

Activations Density 0.263%

phrases indicating negative or undesirable situations

statements that emphasize negation or denial

No Comments

No Known Activations