INDEX

Explanations

statements where something is expected or anticipated

oai_token-act-pair · gpt-3.5-turbo

phrases related to expectations or predictions

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

reen

-0.65

false

-0.65

pmwiki

-0.65

staff

-0.64

 Bach

-0.62

 suggestion

-0.59

 Secrets

-0.59

 Mayo

-0.58

matter

-0.58

clear

-0.58

POSITIVE LOGITS

WARD

0.78

OLOG

0.77

antly

0.76

iour

0.75

olate

0.74

 bruises

0.72

OLOGY

0.72

oval

0.71

ropy

0.69

orr

0.68

Activations Density 0.021%

statements where something is expected or anticipated

phrases related to expectations or predictions

No Comments

No Known Activations