INDEX

Explanations

phrases indicating that something has been well-established or well-known

oai_token-act-pair · gpt-3.5-turbo

the term "well" in various contexts indicating quality or reputation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

hyde

-0.86

rush

-0.77

hip

-0.72

ategory

-0.72

atto

-0.71

ngth

-0.70

omore

-0.67

olean

-0.65

ataka

-0.64

illary

-0.64

POSITIVE LOGITS

enough

1.12

spring

1.09

 enough

1.04

 behaved

1.00

 suited

0.97

 deserved

0.82

served

0.82

 vers

0.82

baum

0.81

 positioned

0.81

Activations Density 0.041%

phrases indicating that something has been well-established or well-known

the term "well" in various contexts indicating quality or reputation

No Comments

No Known Activations