INDEX

Explanations

references to canals

oai_token-act-pair · gpt-3.5-turbo

mentions of specific geographical locations and waterways

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

redients

-0.78

doms

-0.74

ARS

-0.73

tle

-0.72

TOP

-0.71

night

-0.70

 Marketable

-0.70

cash

-0.70

SET

-0.69

---------

-0.69

POSITIVE LOGITS

 Canal

1.17

 canal

1.11

 Tunnel

0.80

 Zone

0.79

 reefs

0.77

 Dock

0.76

obar

0.76

ieri

0.75

agall

0.74

 ejac

0.72

Activations Density 0.011%

references to canals

mentions of specific geographical locations and waterways

No Comments

No Known Activations