vwxyzjn

from collections import defaultdict
import numpy as np
import pandas as pd
import torch
import torch.nn as nn
from datasets import load_dataset
from rich.console import Console
from rich.table import Table
from transformers import (
    AutoConfig,

vwxyzjn

from dataclasses import dataclass
from datasets import load_dataset
import llm_blender
from transformers import HfArgumentParser
import multiprocessing
import random
import itertools
import warnings
from collections import defaultdict
warnings.filterwarnings("ignore")

vwxyzjn

from collections import defaultdict
import numpy as np
import pandas as pd
import torch
import torch.nn as nn
from datasets import load_dataset
from rich.console import Console
from rich.table import Table
from transformers import (
    AutoConfig,

vwxyzjn

import torch
import transformers
import torch.nn.functional as F

tokenizer = transformers.AutoTokenizer.from_pretrained("gpt2", padding_side="right")
tokenizer.add_special_tokens({"pad_token": "[PAD]"})
pad_id = tokenizer.pad_token_id
policy = transformers.AutoModelForCausalLM.from_pretrained("gpt2")
policy.generation_config.pad_token_id = policy.generation_config.eos_token_id

vwxyzjn

# docs and experiment results can be found at https://docs.cleanrl.dev/rl-algorithms/ppo/#ppo_atari_multigpupy
import os
import random
import time
import warnings
from dataclasses import dataclass, field
from typing import List, Literal

import gymnasium as gym
import numpy as np

vwxyzjn

from transformers import AutoTokenizer, AutoModelForCausalLM
import numpy as np
from scipy.special import softmax
model_name = "hkust-nlp/deita-quality-scorer"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

def infer_quality(model, tokenizer, input_text, output_text):
    quality_template = ("You are a helpful assistant. Please identify the quality score of the Response corresponding to the Question. \n #Question#:\n{instruction}\n#Response#:\n{output} \n##Quality: ")
    user_input = quality_template.format(instruction=input_text, output=output_text)

vwxyzjn

(.venv) costa@login-node-1:/fsx/costa/tgi-swarm$ python examples/benchmark.py --instances=1 --model mistralai/Mixtral-8x7B-Instruct-v0.1
None of PyTorch, TensorFlow >= 2.0, or Flax have been found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
running sbatch --parsable slurm/tgi_1707320176_tgi.slurm
Slurm Job ID: ['1774193']
📖 Slurm hosts path: slurm/tgi_1707320176_host_tgi.txt
✅ Done! Waiting for 1774193 to be created                                                                                                           
📖 Slurm log path: slurm/logs/llm-swarm_1774193.out
✅ Done! Waiting for slurm/tgi_1707320176_host_tgi.txt to be created                                                                                 
obtained endpoints ['http://26.0.171.88:25145']
⣽ Waiting for http://26.0.171.88:25145 to be reachable

vwxyzjn

import os
import random
import time
from collections import defaultdict
from dataclasses import asdict, dataclass, field
from types import SimpleNamespace
from typing import List, Literal, Optional

import numpy as np
import pandas as pd

vwxyzjn

from collections import defaultdict
import numpy as np
import pandas as pd
import torch
import torch.nn as nn
from datasets import load_dataset
from rich.console import Console
from rich.table import Table
from transformers import (
    AutoConfig,

vwxyzjn

import torch
import transformers
import torch.nn.functional as F

tokenizer = transformers.AutoTokenizer.from_pretrained("gpt2", padding_side="right")
tokenizer.add_special_tokens({"pad_token": "[PAD]"})
pad_id = tokenizer.pad_token_id
policy = transformers.AutoModelForCausalLM.from_pretrained("gpt2")
policy.generation_config.pad_token_id = policy.generation_config.eos_token_id