INDEX

Explanations

words related to a specific category or theme, particularly with the recurring pattern of "ild" combinations

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "child" or "children"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

SOURCE

-0.81

andise

-0.80

MIT

-0.71

 Yates

-0.69

CHR

-0.68

Mich

-0.68

OURCE

-0.67

displayText

-0.66

perature

-0.65

éŃĶ

-0.64

POSITIVE LOGITS

ild

0.93

erers

0.90

roid

0.82

doms

0.80

itudinal

0.79

nesses

0.77

enstein

0.76

reth

0.76

inen

0.76

erer

0.75

Activations Density 0.007%

words related to a specific category or theme, particularly with the recurring pattern of "ild" combinations

references to the concept of "child" or "children"

No Comments

No Known Activations