INDEX

Explanations

words related to a specific company or source, in this case, "Telegraph."

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "telegraph" and related variations

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

GD

-0.68

 Salvador

-0.61

osit

-0.60

const

-0.59

 Augustine

-0.59

Hut

-0.59

HG

-0.58

walker

-0.58

xxx

-0.58

 rounding

-0.58

POSITIVE LOGITS

legraph

1.23

orld

0.91

shaw

0.77

itudinal

0.76

urst

0.75

hedon

0.75

å¥³

0.75

ipeg

0.74

hester

0.73

 NEWS

0.72

Activations Density 0.009%

words related to a specific company or source, in this case, "Telegraph."

occurrences of the word "telegraph" and related variations

No Comments

No Known Activations