INDEX

Explanations

references to a television network or channel (ND TV)

oai_token-act-pair · gpt-3.5-turbo

mentions of the acronym "ND" and related items, indicating a focus on news topics

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

pell

-0.91

acters

-0.90

hold

-0.83

acter

-0.79

============

-0.78

========

-0.77

rosse

-0.75

================

-0.74

cent

-0.74

chwitz

-0.70

POSITIVE LOGITS

ND

1.06

epend

0.91

ND

0.89

ependent

0.84

ENSE

0.82

LR

0.80

IFF

0.76

NF

0.74

irect

0.74

IA

0.74

Activations Density 0.011%

references to a television network or channel (ND TV)

mentions of the acronym "ND" and related items, indicating a focus on news topics

No Comments

No Known Activations