INDEX

Explanations

phrases indicating consequence or logical reasoning

oai_token-act-pair · gpt-3.5-turbo

conjunctions that imply causality

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

 Rumble

-0.69

 Defenders

-0.69

 Feld

-0.66

Fram

-0.65

MM

-0.64

 Tacoma

-0.63

MM

-0.62

 nurs

-0.60

 Twist

-0.60

 straw

-0.59

POSITIVE LOGITS

forth

1.09

 facto

0.79

ĵĺ

0.75

ettings

0.75

manuel

0.74

uration

0.73

otent

0.72

ptions

0.70

akings

0.70

xual

0.70

Activations Density 0.018%

phrases indicating consequence or logical reasoning

conjunctions that imply causality

No Comments

No Known Activations