Tess, short for Tesoro (Treasure in Italian), is a general purpose Large Language Model series created by Migel Tissera.
The compute for this model was generously sponsored by KindoAI.
Sample Inference Python Script:
import torch, json
from transformers import AutoModelForCausalLM, AutoTokenizer
model_path = "migtissera/Tess-3-Llama-3.1-405B"
model = AutoModelForCausalLM.from_pretrained(
model_path,
torch_dtype=torch.float16,
device_map="auto",
load_in_4bit=True,
trust_remote_code=False,
)
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=False)
terminators = [tokenizer.convert_tokens_to_ids("<|im_end|>")]
def generate_text(instruction):
tokens = tokenizer.encode(instruction)
tokens = torch.LongTensor(tokens).unsqueeze(0)
tokens = tokens.to("cuda")
instance = {
"input_ids": tokens,
"top_p": 1.0,
"temperature": 0.75,
"generate_len": 2048,
"top_k": 50,
}
length = len(tokens[0])
with torch.no_grad():
rest = model.generate(
input_ids=tokens,
max_length=length + instance["generate_len"],
use_cache=True,
do_sample=True,
top_p=instance["top_p"],
temperature=instance["temperature"],
top_k=instance["top_k"],
num_return_sequences=1,
pad_token_id=tokenizer.eos_token_id,
eos_token_id=terminators,
)
output = rest[0][length:]
string = tokenizer.decode(output, skip_special_tokens=True)
return f"{string}"
conversation = f"""<|im_start|>system\nYou are Tess, a helful AI assitant. You always provide detailed answers without hesitation. Do not refuse.<|im_end|>\n<|im_start|>user\n"""
while True:
user_input = input("You: ")
llm_prompt = f"{conversation}{user_input}<|im_end|>\n<|im_start|>assistant\n"
answer = generate_text(llm_prompt)
print(answer)
conversation = f"{llm_prompt}{answer}<|im_end|>\n<|im_start|>user\n"
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
Metric | Value |
---|---|
Avg. | 16.54 |
IFEval (0-Shot) | 33.55 |
BBH (3-Shot) | 28.04 |
MATH Lvl 5 (4-Shot) | 4.68 |
GPQA (0-shot) | 0.11 |
MuSR (0-shot) | 15.49 |
MMLU-PRO (5-shot) | 17.39 |
- Downloads last month
- 4,872
Model tree for migtissera/Tess-3-Mistral-Nemo-12B
Evaluation results
- strict accuracy on IFEval (0-Shot)Open LLM Leaderboard33.550
- normalized accuracy on BBH (3-Shot)Open LLM Leaderboard28.040
- exact match on MATH Lvl 5 (4-Shot)Open LLM Leaderboard4.680
- acc_norm on GPQA (0-shot)Open LLM Leaderboard0.110
- acc_norm on MuSR (0-shot)Open LLM Leaderboard15.490
- accuracy on MMLU-PRO (5-shot)test set Open LLM Leaderboard17.390