INDEX

Explanations

the word "feared" or variations of it, indicating a sense of apprehension or concern

oai_token-act-pair · gpt-3.5-turbo

expressions of concern or apprehension

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

estone

-0.82

ysis

-0.78

etry

-0.77

oard

-0.76

ithing

-0.76

vertis

-0.75

emark

-0.74

etts

-0.73

ricks

-0.73

ession

-0.73

POSITIVE LOGITS

lessly

0.90

 retribution

0.82

 lest

0.80

ILY

0.75

 fate

0.74

ROR

0.74

 fears

0.68

 persecution

0.68

 complicity

0.68

 retaliation

0.68

Activations Density 0.038%

the word "feared" or variations of it, indicating a sense of apprehension or concern

expressions of concern or apprehension

No Comments

No Known Activations