INDEX

Explanations

phrases related to specific organizations or institutions such as companies, news agencies, and foundations

oai_token-act-pair · gpt-3.5-turbo

the repeated sequence "OM" across various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

wall

-0.79

hs

-0.71

wright

-0.68

place

-0.67

yards

-0.67

waters

-0.66

lift

-0.66

 whatever

-0.63

 otherwise

-0.63

flags

-0.63

POSITIVE LOGITS

obile

1.05

orrow

1.04

EGA

1.03

OM

1.02

atis

0.96

edia

0.96

ondo

0.91

atoes

0.91

ulus

0.91

orph

0.90

Activations Density 0.006%

phrases related to specific organizations or institutions such as companies, news agencies, and foundations

the repeated sequence "OM" across various contexts

No Comments

No Known Activations