INDEX

Explanations

adjectives describing something altered, distorted, or perverted

oai_token-act-pair · gpt-3.5-turbo

terms related to distortion and manipulation of reality

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

phabet

-0.78

buy

-0.78

-0.76

idates

-0.73

Aud

-0.73

ingu

-0.73

meet

-0.72

availability

-0.67

 Nationals

-0.67

 acquisition

-0.66

POSITIVE LOGITS

 twisted

3.62

 warped

2.26

 distorted

1.95

 Twisted

1.86

 twisting

1.80

 tangled

1.61

 skewed

1.43

 mang

1.39

 spun

1.38

 perverse

1.35

Activations Density 0.020%

adjectives describing something altered, distorted, or perverted

terms related to distortion and manipulation of reality

No Comments

No Known Activations