INDEX

Explanations

code-related text patterns such as module-specific phrases and variable setters

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

)."

-0.25

 ......

-0.22

 [...]

-0.22

 tremend

-0.21

 âĸĪ

-0.21

!".

-0.21

"""

-0.20

)",

-0.20

↵Âł

-0.20

 ÂłÂł

-0.20

POSITIVE LOGITS

ifting

0.17

rug

0.17

ooked

0.17

Preview

0.17

 optimism

0.16

atform

0.16

ruce

0.16

urring

0.16

 ÂŃ

0.16

ancing

0.16

Activations Density 5.442%

code-related text patterns such as module-specific phrases and variable setters

No Comments

No Known Activations