INDEX

Explanations

mentions of the word "wired"

oai_token-act-pair · gpt-3.5-turbo

words related to fatigue or exhaustion

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

 ç¥ŀ

-0.72

Bei

-0.67

oulos

-0.66

 amen

-0.65

MER

-0.61

 Bundy

-0.61

 Barcl

-0.61

romeda

-0.60

Ń·

-0.59

alter

-0.59

POSITIVE LOGITS

lessly

0.94

rums

0.85

ilage

0.83

anto

0.82

ansas

0.80

rum

0.80

itably

0.80

rack

0.79

ndra

0.77

dit

0.76

Activations Density 0.020%

mentions of the word "wired"

words related to fatigue or exhaustion

No Comments

No Known Activations