INDEX

Explanations

references to the word "Beer."

oai_token-act-pair · gpt-3.5-turbo

references to beer

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

uncture

-0.96

iculty

-0.75

itives

-0.73

resses

-0.73

ean

-0.72

ktop

-0.70

etheless

-0.69

urities

-0.69

SPONSORED

-0.69

eanor

-0.68

POSITIVE LOGITS

 Beer

1.11

Beer

0.93

bell

0.77

beer

0.73

Gim

0.73

 Barrel

0.72

wine

0.72

haus

0.71

leigh

0.70

 Lover

0.69

Activations Density 0.009%

references to the word "Beer."

references to beer

No Comments

No Known Activations