liminerity/Bitnet-Mistral.0.2-70m

"""this is my first attempt at converting a model float16 quantized model to 1.5bit. i used alpindale/Mistral-7B-v0.2-hf for the base model and \n trained on: abideen/cosmopedia-100k-pretain dataset and used his google colab project to make this"""

#EXAMPLE INFERENCE CODE FROM ABIDEEN'S COLAB PROJECT

from transformers import AutoModelForCausalLM, AutoTokenizer
from transformers.models.llama.modeling_llama import *
# Load a pretrained BitNet model
model = "liminerity/Bitnet-Mistral.0.2-70M"
tokenizer = AutoTokenizer.from_pretrained(model)
model = AutoModelForCausalLM.from_pretrained(model)


def activation_quant(x):
    scale = 127.0 / x.abs().max(dim=-1, keepdim=True).values.clamp_(min=1e-5)
    y = (x * scale).round().clamp_(-128, 127)
    y = y / scale
    return y
def weight_quant(w):
    scale = 1.0 / w.abs().mean().clamp_(min=1e-5)
    u = (w * scale).round().clamp_(-1, 1)
    u = u / scale
    return u

class BitLinear(nn.Linear):
    def forward(self, x):
        w = self.weight # a weight tensor with shape [d, k]
        x = x.to(w.device)
        RMSNorm = LlamaRMSNorm(x.shape[-1]).to(w.device)
        x_norm = RMSNorm(x)
        # A trick for implementing Straight−Through−Estimator (STE) using detach()
        x_quant = x_norm + (activation_quant(x_norm) - x_norm).detach()
        w_quant = w + (weight_quant(w) - w).detach()
        y = F.linear(x_quant, w_quant)
        return y

def convert_to_bitnet(model, copy_weights):
    for name, module in model.named_modules():
        # Replace linear layers with BitNet
        if isinstance(module, LlamaSdpaAttention) or isinstance(module, LlamaMLP):
            for child_name, child_module in module.named_children():
                if isinstance(child_module, nn.Linear):
                    bitlinear = BitLinear(child_module.in_features, child_module.out_features, child_module.bias is not None).to(device="cuda:0")
                    if copy_weights:
                        bitlinear.weight = child_module.weight
                        if child_module.bias is not None:
                            bitlinear.bias = child_module.bias
                    setattr(module, child_name, bitlinear)
        # Remove redundant input_layernorms
        elif isinstance(module, LlamaDecoderLayer):
            for child_name, child_module in module.named_children():
                if isinstance(child_module, LlamaRMSNorm) and child_name == "input_layernorm":
                    setattr(module, child_name, nn.Identity().to(device="cuda:0"))


convert_to_bitnet(model, copy_weights=True)
model.to(device="cuda:0")

prompt = "What is Machine Learning?"
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
generate_ids = model.generate(inputs.input_ids, max_length=50)
tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]