INDEX

Explanations

references to a specific location called "The Hill."

oai_token-act-pair · gpt-3.5-turbo

mentions of "Hill" in various contexts

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

glomer

-0.73

ULAR

-0.71

============

-0.70

 oppos

-0.69

 Alonso

-0.67

cial

-0.66

uality

-0.65

detail

-0.65

========

-0.63

 Bucc

-0.63

POSITIVE LOGITS

iard

1.33

yer

1.08

top

0.93

side

0.90

castle

0.90

shire

0.89

hog

0.89

marks

0.87

bill

0.87

boro

0.86

Activations Density 0.021%

references to a specific location called "The Hill."

mentions of "Hill" in various contexts

No Comments

No Known Activations