INDEX

Explanations

names of places and people

oai_token-act-pair · gpt-3.5-turbo

references to organizations or institutions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

Newsletter

-0.76

 Trees

-0.71

 Tsukuyomi

-0.70

 Vader

-0.69

 Seller

-0.69

↵Âł

-0.69

camp

-0.69

éĹĺ

-0.69

 Gerrard

-0.68

CONCLUS

-0.67

POSITIVE LOGITS

 embattled

0.90

attled

0.77

 controversial

0.71

 newly

0.69

 hotly

0.69

 defiant

0.66

 overdue

0.65

UST

0.64

awaited

0.64

 blockbuster

0.64

Activations Density 0.983%

names of places and people

references to organizations or institutions

No Comments

No Known Activations