INDEX

Explanations

names, likely related to individuals or companies

oai_token-act-pair · gpt-3.5-turbo

terms related to legal or judicial processes

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

 reservations

-0.69

 fried

-0.69

 sauces

-0.69

nant

-0.69

 sshd

-0.68

elve

-0.67

 restaurants

-0.66

 Hungry

-0.65

iston

-0.65

 reservation

-0.65

POSITIVE LOGITS

 Spiegel

0.98

odox

0.79

âĸ¬

0.77

âĹ¼

0.77

Newsletter

0.75

channelAvailability

0.74

Rah

0.73

scribed

0.73

board

0.72

onym

0.72

Activations Density 0.048%

names, likely related to individuals or companies

terms related to legal or judicial processes

No Comments

No Known Activations