INDEX

Explanations

mentions of tools or actions related to hooking or attaching objects

oai_token-act-pair · gpt-3.5-turbo

references to "hooks" 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 @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

uria

-0.93

 Shinra

-0.84

PLA

-0.81

adena

-0.75

intosh

-0.71

erva

-0.71

 Catalonia

-0.70

abe

-0.69

usat

-0.69

SAP

-0.68

POSITIVE LOGITS

 hook

4.03

 hooks

2.79

hook

2.57

 Hook

2.53

 hooked

1.86

 hitch

1.13

 loop

1.05

 tether

1.02

 bait

1.00

 strap

0.98

Activations Density 0.017%

mentions of tools or actions related to hooking or attaching objects

references to "hooks" in various contexts

No Comments

No Known Activations