INDEX

Explanations

references to records or recording in a text

oai_token-act-pair · gpt-3.5-turbo

occurrences of the term "rec."

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

ahime

-0.74

MENTS

-0.72

furt

-0.72

 Tasman

-0.72

tips

-0.72

 Panzer

-0.66

falls

-0.65

Magikarp

-0.65

 Kenyan

-0.64

loo

-0.63

POSITIVE LOGITS

ycle

1.29

urrent

1.12

ruit

1.11

ruits

1.10

ipient

1.09

ipes

1.06

ording

1.05

ords

1.03

oded

1.01

onnaissance

0.98

Activations Density 0.012%

references to records or recording in a text

occurrences of the term "rec."

No Comments

No Known Activations