rombo dawg

rombodawg

AI & ML interests

Datasets, Finetuning ㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤ ‎‎‎‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‎‎‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‎‎‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‎‎‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‎‎‎‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ ‎‏ Please check out my Org: https://huggingface.co/Replete-AI and my new dataset at Replete-AI/code_bagel

Organizations

rombodawg's activity

New activity in jondurbin/bagel-v0.5 1 day ago

I have created the other half of your bagel 😋

#2 opened 1 day ago by

rombo dawg

AI & ML interests

Organizations

rombodawg's activity

I have created the other half of your bagel 😋

Update README.md

Thank you for making this!

[bot] Conversion to Parquet

Context windows is only 8k???

correct license

Passes "snake in python test" after only 3 attempts

Does the model actually work?

ITS NOT REAL

ITS NOT REAL

C_H_U_N_K_Y-L_L_A_M_A

Very good model for its size (7b and 14b are gonna be amazing)

What is 256k?

Mind adding description for this dataset?

Good model but Bullshit chart and inaccurate numbers

Openchat-llama-3-8b releasing soon! Please train another CodeNinja OpencodeInterpreter data (linked)

Fix name from 8b to 70b

Not at all uncensored

Can you train llama-3-8b-instruct? next

failed run Llama-3-11.5B-v2

Bad example image

I made a similar dataset with more code if you are interested

What merge method was used?

What kind of dataset?

Mistral-evolved-11b version??

Failed run

What datasets were these trained on?

help me understand the point of such scaling

FP16 files created for your convenience

Can code snake in python, first model at this size after deekseekcoder-instruct-6.7b (With caveat)

(Rebuked: this claim proven false) "Fake coding scores" .73 at best

is 14b coming?

You should do this with Gemma-7b

is this a single expert of all the experts merged?

We are working on creating a single 22b from this model

7B or 8B?

Are you gonna finetune the code version of this model on your own dataset?

Thank you for posting this

This model card is a blessing

Number of parameters

Why is this completely broken?

Fine tune on top of the new stable code model

Train Mistral 7B 0.2

What was the order of training?

what kind of dataset is Tess?

Update README.md

How did you increase the size to 11b?

Update README.md

Coding performance of base model?

Hey thanks for using my model 😉

I submitted my model a while ago and it never got benchmarked

Question

Create model card

I made the next gen mistral model. Id love to see you train it in the same way!

Submit Mistral-Evolved-11b-v0.1 to the HuggingFaceH4 Open LLM LeaderBoard

Seperate larger from smaller model benchmarking.

Try my model next

Add `chat_template` in tokenizer config

Please create google Gemma-7b (8.5b) based version

Update README.md