INDEX

Explanations

words related to negative or critical contexts

oai_token-act-pair · gpt-3.5-turbo

words or terms related to surprise, anger, and various structured themes or events

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 âĨĳ

-0.70

 Leilan

-0.68

 ---------

-0.67

 Naples

-0.64

 ãĢĮ

-0.64

 Eugene

-0.64

 Florence

-0.63

UMP

-0.61

FN

-0.61

å¿

-0.60

POSITIVE LOGITS

1.10

"],

1.09

"]

1.01

"!

1.00

",

0.98

":

0.97

tainment

0.96

"âĢ¦

0.94

")

0.94

"-

0.93

Activations Density 0.225%

words related to negative or critical contexts

words or terms related to surprise, anger, and various structured themes or events

No Comments

No Known Activations