INDEX

Explanations

the word 'to' in various contexts

oai_token-act-pair · gpt-3.5-turbo

phrases that include "I like to" followed by various verbs

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

VERTISEMENT

-0.74

Required

-0.74

grounds

-0.68

Abs

-0.68

lights

-0.65

proof

-0.64

Reviewer

-0.63

soDeliveryDate

-0.61

KS

-0.61

packing

-0.60

POSITIVE LOGITS

 emulate

1.04

 remind

1.03

 propose

1.02

 dispose

0.99

 capitalize

0.97

 encourage

0.94

 emphasize

0.92

 recreate

0.92

 commemorate

0.92

 partake

0.92

Activations Density 0.085%

the word 'to' in various contexts

phrases that include "I like to" followed by various verbs

No Comments

No Known Activations