hanzoai/open-instruct
2
0
Fork 0
mirror of https://github.com/hanzoai/open-instruct.git synced 2026-10-30 18:26:32 +00:00
Code Issues Packages Projects Releases Wiki Activity
Files
rescue/head-9d3e8a4056
open-instruct/scripts/synth_pref/utils/model_configs.py
T
Lj MirandaandGitHub a5fe119f7d Add code for synthetic preference pipeline (#442) 
* Add .vscode to .gitignore

* Add dependencies for synth pref pipeline

* Initialize directory

* Port majority of the scripts to open-instruct

* Add documentation

* Add public birr as a submodule

* Minor fixes and update README

* Add instructions for creating annotation mix

* Run isort on source

* Update README

* Remove birr submodule

* Change birr from submodule to git-checkout

* Add examples

* Update README with better examples
2024-11-21 07:28:06 -08:00

257 lines
7.7 KiB
Python
Raw Permalink Blame History

from typing import Any
default_generation_params = {
"temperature": 1,
"top_p": 1,
"max_tokens": 7500,
}
MODELS: dict[str, dict[str, Any]] = {
"Meta-Llama/Llama-3-8B-Instruct": {
"model": {
"name_or_path": "Meta-Llama/Llama-3-8B-Instruct",
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 1},
},
"Meta-Llama/Llama-3.1-8B-Instruct": {
"model": {
"name_or_path": "Meta-Llama/Llama-3.1-8B-Instruct",
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 1},
},
"Meta-Llama/Llama-3.1-70B-Instruct": {
"model": {
"name_or_path": "Meta-Llama/Llama-3.1-70B-Instruct",
"tensor_parallel_size": 4,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 4},
},
"Meta-Llama/Llama-3-405B-Instruct-fp8": {
"model": {
"name_or_path": "Meta-Llama/Llama-3-405B-Instruct-fp8",
"tensor_parallel_size": 8,
"num_scheduler_steps": 1,
"max_model_len": 64000,
},
"generate": default_generation_params,
"task": {"gpu_count": 8},
},
"mistralai/Mistral-7B-Instruct-v0.2": {
"model": {
"name_or_path": "mistralai/Mistral-7B-Instruct-v0.2",
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 1},
},
"mistralai/Mistral-Nemo-Instruct-2407": {
"model": {
"name_or_path": "mistralai/Mistral-Nemo-Instruct-2407",
"tensor_parallel_size": 2,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 2},
},
"allenai/tulu-2-7b": {
"model": {
"name_or_path": "allenai/tulu-2-7b",
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 1},
},
"allenai/tulu-2-13b": {
"model": {
"name_or_path": "allenai/tulu-2-13b",
"tensor_parallel_size": 2,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 2},
},
"allenai/tulu-2-70b": {
"model": {
"name_or_path": "allenai/tulu-2-70b",
"tensor_parallel_size": 4,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 4},
},
"01-ai/Yi-34B-Chat": {
"model": {
"name_or_path": "01-ai/Yi-34B-Chat",
"tensor_parallel_size": 2,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 2},
},
"01-ai/Yi-6B-Chat": {
"model": {
"name_or_path": "01-ai/Yi-6B-Chat",
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 1},
},
"allenai/tulu-2-7b": {
"model": {
"name_or_path": "allenai/tulu-2-7b",
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 1},
},
"mosaicml/mpt-30b-chat": {
"model": {
"name_or_path": "mosaicml/mpt-30b-chat",
"tensor_parallel_size": 2,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 2},
},
"mosaicml/mpt-7b-8k-chat": {
"model": {
"name_or_path": "mosaicml/mpt-7b-8k-chat",
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 1},
},
"google/gemma-2-27b-it": {
"model": {
"name_or_path": "google/gemma-2-27b-it",
"tensor_parallel_size": 2,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 2},
},
"google/gemma-2-9b-it": {
"model": {
"name_or_path": "google/gemma-2-9b-it",
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 1},
},
"internlm/internlm2_5-20b-chat": {
"model": {
"name_or_path": "internlm/internlm2_5-20b-chat",
"trust_remote_code": True,
"tensor_parallel_size": 2,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 2},
},
"internlm/internlm2_5-7b-chat": {
"model": {
"name_or_path": "internlm/internlm2_5-7b-chat",
"trust_remote_code": True,
"tensor_parallel_size": 2,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 2},
},
"internlm/internlm2_5-1_8b-chat": {
"model": {
"name_or_path": "internlm/internlm2_5-1_8b-chat",
"trust_remote_code": True,
"tensor_parallel_size": 2,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 2},
},
"tiiuae/falcon-7b-instruct": {
"model": {
"name_or_path": "tiiuae/falcon-7b-instruct",
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 1},
},
"Qwen/Qwen2.5-72B-Instruct": {
"model": {
"name_or_path": "Qwen/Qwen2.5-72B-Instruct",
"tensor_parallel_size": 4,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 4},
},
"Qwen/Qwen2.5-32B-Instruct": {
"model": {
"name_or_path": "Qwen/Qwen2.5-32B-Instruct",
"tensor_parallel_size": 2,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 2},
},
"Qwen/Qwen2.5-14B-Instruct": {
"model": {
"name_or_path": "Qwen/Qwen2.5-14B-Instruct",
"tensor_parallel_size": 2,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 2},
},
"Qwen/Qwen2.5-7B-Instruct": {
"model": {
"name_or_path": "Qwen/Qwen2.5-7B-Instruct",
"tensor_parallel_size": 2,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 2},
},
"Qwen/Qwen2.5-3B-Instruct": {
"model": {
"name_or_path": "Qwen/Qwen2.5-3B-Instruct",
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 1},
},
"allenai/llama-tulu-3-70b": {
"model": {
"name_or_path": "allenai/llama-tulu-3-70b",
"tensor_parallel_size": 4,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 4},
},
"allenai/llama-tulu-3-8b": {
"model": {
"name_or_path": "allenai/llama-tulu-3-8b",
"tensor_parallel_size": 2,
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 2},
},
"allenai/OLMo-7B-0724-SFT-hf": {
"model": {
"name_or_path": "allenai/OLMo-7B-0724-SFT-hf",
"num_scheduler_steps": 1,
},
"generate": default_generation_params,
"task": {"gpu_count": 1},
},
}
Reference in New Issue View Git Blame Copy Permalink