Libre AI Models #12

Open
opened 2023-11-15 09:21:00 -07:00 by rs-parrot · 19 comments
rs-parrot commented 2023-11-15 09:21:00 -07:00 (Migrated from spacecruft.org)

Determine which are the SOTA libre models at various scales.

See:

https://spacecruft.org/deepcrayon/parrot-models

Determine which are the SOTA libre models at various scales. See: https://spacecruft.org/deepcrayon/parrot-models
rs-parrot commented 2023-11-15 09:21:08 -07:00 (Migrated from spacecruft.org)

Check which of these models used in Ollama have good licenses:

codellama:34b-code
codellama:34b-code-q8_0
codellama:34b-instruct-q8_0
codellama:34b-python
codellama:34b-python-q8_0
everythinglm:13b-16k-q8_0
falcon:40b
falcon:40b-instruct
falcon:40b-instruct-fp16
falcon:40b-text
falcon:40b-text-fp16
falcon:180b
falcon:180b-chat-q4_0
falcon:180b-text-q4_0
falcon:180b-text-q8_0
llama2:70b
mistral:7b-instruct-q8_0
mistral:7b-text-q8_0
nexusraven:13b
nexusraven:13b-q8_0
orca-mini:latest
phind-codellama:34b-python
phind-codellama:34b-python-q8_0 9e8121c0c614 36
phind-codellama:34b-v2
phind-codellama:34b-v2-q2_K
phind-codellama:34b-v2-q4_0
phind-codellama:34b-v2-q4_1
phind-codellama:34b-v2-q4_K_S
phind-codellama:34b-v2-q6_K
phind-codellama:34b-v2-q8_0
sqlcoder:15b
sqlcoder:15b-fp16
sqlcoder:15b-q8_0
starcoder:15b-base-q8_0
vicuna:33b
vicuna:33b-q8_0
wizardcoder:7b-python
wizardcoder:13b-python
wizardcoder:34b-python
wizardcoder:34b-python-q2_K
wizardcoder:34b-python-q6_K
wizardcoder:34b-python-q8_0
wizard-math:70b
wizard-math:70b-q5_K_M

Check which of these models used in Ollama have good licenses: codellama:34b-code codellama:34b-code-q8_0 codellama:34b-instruct-q8_0 codellama:34b-python codellama:34b-python-q8_0 everythinglm:13b-16k-q8_0 falcon:40b falcon:40b-instruct falcon:40b-instruct-fp16 falcon:40b-text falcon:40b-text-fp16 falcon:180b falcon:180b-chat-q4_0 falcon:180b-text-q4_0 falcon:180b-text-q8_0 llama2:70b mistral:7b-instruct-q8_0 mistral:7b-text-q8_0 nexusraven:13b nexusraven:13b-q8_0 orca-mini:latest phind-codellama:34b-python phind-codellama:34b-python-q8_0 9e8121c0c614 36 phind-codellama:34b-v2 phind-codellama:34b-v2-q2_K phind-codellama:34b-v2-q4_0 phind-codellama:34b-v2-q4_1 phind-codellama:34b-v2-q4_K_S phind-codellama:34b-v2-q6_K phind-codellama:34b-v2-q8_0 sqlcoder:15b sqlcoder:15b-fp16 sqlcoder:15b-q8_0 starcoder:15b-base-q8_0 vicuna:33b vicuna:33b-q8_0 wizardcoder:7b-python wizardcoder:13b-python wizardcoder:34b-python wizardcoder:34b-python-q2_K wizardcoder:34b-python-q6_K wizardcoder:34b-python-q8_0 wizard-math:70b wizard-math:70b-q5_K_M
rs-parrot commented 2023-11-15 09:21:17 -07:00 (Migrated from spacecruft.org)

Llama's license isn't libre.

Llama's license isn't libre.
rs-parrot commented 2023-11-15 09:21:22 -07:00 (Migrated from spacecruft.org)

Llama non-free license includes these:

NexusRaven
Codellama
Phind-CodeLlama
WizardCoder
Vicuna
WizardLM
CodeLlama

Llama non-free license includes these: NexusRaven Codellama Phind-CodeLlama WizardCoder Vicuna WizardLM CodeLlama
rs-parrot commented 2023-11-15 09:21:27 -07:00 (Migrated from spacecruft.org)

BigCode OpenRAIL-M and OpenRAIL++ are non-free licenses, includes these:

CodeUp-Llama
StarCoder
WizardCoder-15B-V1.0

BigCode OpenRAIL-M and OpenRAIL++ are non-free licenses, includes these: CodeUp-Llama StarCoder WizardCoder-15B-V1.0
rs-parrot commented 2023-11-15 09:21:32 -07:00 (Migrated from spacecruft.org)

Other proprietary licenses:

Claude
GPT
PaLM
Qwen

StableCode Research License is non-free:
StableCode

Other proprietary licenses: Claude GPT PaLM Qwen StableCode Research License is non-free: StableCode
rs-parrot commented 2023-11-15 09:21:38 -07:00 (Migrated from spacecruft.org)

Possibly libre, under apache 2 license:

Falcon
Mistral
Pythia

Possibly libre, under CC-BY-SA:

SQLCoder
Stable LM
Open-Assistant StableLM

Possibly libre, under apache 2 license: Falcon Mistral Pythia Possibly libre, under CC-BY-SA: SQLCoder Stable LM Open-Assistant StableLM
rs-parrot commented 2023-11-15 09:21:43 -07:00 (Migrated from spacecruft.org)

Phind 34b v2 works well with llama.cpp and parrot with continuedev extension. It has fake apache license that makes it non-libre.

Phind 34b v2 works well with llama.cpp and parrot with continuedev extension. It has fake apache license that makes it non-libre.
rs-parrot commented 2023-11-15 09:22:01 -07:00 (Migrated from spacecruft.org)

Deepseek coder is new, afaict. 33b under MIT, trained on code.

Deepseek coder is new, afaict. 33b under MIT, trained on code.
rs-parrot commented 2023-11-15 09:22:10 -07:00 (Migrated from spacecruft.org)

Meh, deepseek says MIT, but has a fake open source license for the model.

Meh, deepseek says MIT, but has a fake open source license for the model.
rs-parrot commented 2023-11-15 14:10:31 -07:00 (Migrated from spacecruft.org)

This may be actual "open source":

https://github.com/EleutherAI
https://www.eleuther.ai

I don't see a coding model, but it has many others, and tools, which perhaps could make one.

This may be actual "open source": https://github.com/EleutherAI https://www.eleuther.ai I don't see a coding model, but it has many others, and tools, which perhaps could make one.
rs-parrot commented 2023-11-15 14:10:35 -07:00 (Migrated from spacecruft.org)

EleutherAI trains on "The Pile"

https://pile.eleuther.ai/

That full dataset is ~850GB, apache license, available via torrent, etc.

EleutherAI trains on "The Pile" https://pile.eleuther.ai/ That full dataset is ~850GB, apache license, available via torrent, etc.
rs-parrot commented 2023-11-15 14:10:41 -07:00 (Migrated from spacecruft.org)

"Older" model (2022), but appears to actually be libre model (apache 2.0 license, without BS):

https://beta.the-eye.eu/public/AI/models/GPT-NeoX-20B/

"Older" model (2022), but appears to actually be libre model (apache 2.0 license, without BS): https://beta.the-eye.eu/public/AI/models/GPT-NeoX-20B/
rs-parrot commented 2023-11-15 14:10:46 -07:00 (Migrated from spacecruft.org)

Eleuther AI claims "open source", and some parts may be, but it also uses proprietary models, such as llama license, eg.

https://huggingface.co/EleutherAI/llemma_7b

So they aren't all "open source" either, need to double check anything licensed from them.

Eleuther AI claims "open source", and some parts may be, but it also uses proprietary models, such as llama license, eg. https://huggingface.co/EleutherAI/llemma_7b So they aren't all "open source" either, need to double check anything licensed from them.
rs-parrot commented 2023-11-21 09:35:50 -07:00 (Migrated from spacecruft.org)

Coqui license is non-libre.

Coqui license is non-libre.
ate-parrot commented 2023-11-25 18:20:09 -07:00 (Migrated from spacecruft.org)

chatglm3-6b license:

"You will not use the Software for any act that may undermine China's national security and national unity"

chatglm3-6b license: "You will not use the Software for any act that may undermine China's national security and national unity"
rs-parrot commented 2023-11-25 18:42:41 -07:00 (Migrated from spacecruft.org)

Another "open source" license that isn't open source:

Yi Series Models Community License Agreement

Another "open source" license that isn't open source: Yi Series Models Community License Agreement
rs-weblate commented 2023-12-04 15:13:15 -07:00 (Migrated from spacecruft.org)
Maybe https://huggingface.co/TigerResearch
rs-parrot commented 2024-05-09 08:14:53 -06:00 (Migrated from spacecruft.org)

In the last few months there have been many new models released and along with that, more license proliferation.

The following models have been tested and used with Parrot, on an AMD CPU server with six AMD Radeon RX 7900 XTX (tinyrocs). Most have also been tested on Nvidia A16s and/or Nvidia A6000s. Using llama.cpp server with GGUF files, either from upstream, TheBloke, or custom built.

Nous Research Llama 3 70B Instruct
Qwen CodeQwen1.5 7B Chat
Qwen1.5 110B Chat
CodeLlama 70B Instruct
Mistral 7B v0.1
Phind CodeLlama 34B v2
WizardCoder Python 34B v1.0
Deepseek Coder 33B Instruct
Mixtral 8x7B Instruct
OpenCodeInterpreter DS 33B
StarCoder2 3B
Falcon 40B Instruct (on Nvidia?)

CodeQwen is by far the fastest, and is the highest rated on the huggingface code leaderboard for Python. It is from Alibaba and under a unique Alibaba license (e.g. corporate fake open source license).

In the last few months there have been many new models released and along with that, more license proliferation. The following models have been tested and used with Parrot, on an AMD CPU server with six AMD Radeon RX 7900 XTX (tinyrocs). Most have also been tested on Nvidia A16s and/or Nvidia A6000s. Using llama.cpp server with GGUF files, either from upstream, TheBloke, or custom built. Nous Research Llama 3 70B Instruct Qwen CodeQwen1.5 7B Chat Qwen1.5 110B Chat CodeLlama 70B Instruct Mistral 7B v0.1 Phind CodeLlama 34B v2 WizardCoder Python 34B v1.0 Deepseek Coder 33B Instruct Mixtral 8x7B Instruct OpenCodeInterpreter DS 33B StarCoder2 3B Falcon 40B Instruct (on Nvidia?) CodeQwen is by far the fastest, and is the highest rated on the huggingface code leaderboard for Python. It is from Alibaba and under a unique Alibaba license (e.g. corporate fake open source license).
rs-parrot commented 2024-05-09 08:20:28 -06:00 (Migrated from spacecruft.org)

(Pseudo-) open source models recommended by continuedev:

Chat (e.g. code generation):
llama3 70b, 8b
deepseek-coder 33b, 6.7b

Autocomplete:
deepseek-coder 6.7b, 1.3b
StarCoder2 7b, 3b

Embeddings:
transformers.js
nomic-embed-text

(Pseudo-) open source models recommended by continuedev: Chat (e.g. code generation): llama3 70b, 8b deepseek-coder 33b, 6.7b Autocomplete: deepseek-coder 6.7b, 1.3b StarCoder2 7b, 3b Embeddings: transformers.js nomic-embed-text
Sign in to join this conversation.
No labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
deepcrayon/parrot#12
No description provided.