INDEX

Explanations

mentions of winning or achievements

oai_token-act-pair · gpt-3.5-turbo

instances of the word "win."

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

erity

-0.75

umn

-0.64

ikk

-0.64

 protr

-0.62

 includ

-0.62

Uz

-0.61

 footprint

-0.60

Else

-0.60

ciplinary

-0.60

 assembled

-0.59

POSITIVE LOGITS

ners

0.91

nings

0.87

now

0.76

win

0.76

throp

0.75

ception

0.74

iors

0.74

iem

0.74

ces

0.72

't

0.72

Activations Density 0.026%

mentions of winning or achievements

instances of the word "win."

No Comments

No Known Activations