candle/candle-examples/examples/quantized/main.rs

#![allow(dead_code)]
#[cfg(feature = "mkl")]
extern crate intel_mkl_src;

#[cfg(feature = "accelerate")]
extern crate accelerate_src;

use clap::{Parser, ValueEnum};
use std::collections::HashMap;
use std::io::Write;
use tokenizers::Tokenizer;

use candle::quantized::ggml_file::Content;
use candle::quantized::QTensor;
use candle::{DType, Device, IndexOp, Result, Tensor, D};
use candle_nn::{Embedding, Module};
use candle_transformers::generation::LogitsProcessor;

const MAX_SEQ_LEN: usize = 4096;
const DEFAULT_PROMPT: &str = "My favorite theorem is ";

struct RmsNorm {
    inner: candle_nn::LayerNorm,
    span: tracing::Span,
}

impl RmsNorm {
    fn new(scale: QTensor) -> Result<Self> {
        let span = tracing::span!(tracing::Level::TRACE, "rms-norm");
        let scale = scale.dequantize(&Device::Cpu)?;
        let inner = candle_nn::LayerNorm::rms_norm(scale, 1e-5);
        Ok(Self { inner, span })
    }

    fn forward(&self, x: &Tensor) -> Result<Tensor> {
        let _enter = self.span.enter();
        self.inner.forward(x)
    }
}

// QMatMul wrapper adding some tracing.
struct QMatMul {
    inner: candle::quantized::QMatMul,
    span: tracing::Span,
}

impl QMatMul {
    fn from_qtensor(qtensor: QTensor) -> Self {
        let inner = candle::quantized::QMatMul::from_qtensor(qtensor);
        let span = tracing::span!(tracing::Level::TRACE, "qmatmul");
        Self { inner, span }
    }

    fn forward(&self, xs: &Tensor) -> Result<Tensor> {
        let _enter = self.span.enter();
        self.inner.forward(xs)
    }
}

struct LayerWeights {
    attention_wq: QMatMul,
    attention_wk: QMatMul,
    attention_wv: QMatMul,
    attention_wo: QMatMul,
    attention_norm: RmsNorm,
    feed_forward_w1: QMatMul,
    feed_forward_w2: QMatMul,
    feed_forward_w3: QMatMul,
    ffn_norm: RmsNorm,
    n_head: usize,
    n_kv_head: usize,
    head_dim: usize,
    cos: Tensor,
    sin: Tensor,
    kv_cache: Option<(Tensor, Tensor)>,
    span_attn: tracing::Span,
    span_rot: tracing::Span,
    span_mlp: tracing::Span,
}

fn masked_fill(on_false: &Tensor, mask: &Tensor, on_true: f32) -> Result<Tensor> {
    let shape = mask.shape();
    let on_true = Tensor::new(on_true, on_false.device())?.broadcast_as(shape.dims())?;
    let m = mask.where_cond(&on_true, on_false)?;
    Ok(m)
}

impl LayerWeights {
    fn apply_rotary_emb(&self, x: &Tensor, index_pos: usize) -> Result<Tensor> {
        let _enter = self.span_rot.enter();
        let (b_sz, n_head, seq_len, n_embd) = x.dims4()?;
        let cos = self
            .cos
            .narrow(0, index_pos, seq_len)?
            .reshape((seq_len, n_embd / 2, 1))?;
        let sin = self
            .sin
            .narrow(0, index_pos, seq_len)?
            .reshape((seq_len, n_embd / 2, 1))?;
        let cos = cos.broadcast_as((b_sz, 1, seq_len, n_embd / 2, 1))?;
        let sin = sin.broadcast_as((b_sz, 1, seq_len, n_embd / 2, 1))?;
        // This mimics the llama.cpp behavior.
        // https://github.com/ggerganov/llama.cpp/blob/1f0bccb27929e261744c979bc75114955da49e98/ggml.c#L12104-L12105
        // The x0 and x1 value are interleaved on the n_embd (= head_dim) dimension.
        // The resulting y0 and y1 are also interleaved with:
        //   y0 = x0*cos - x1*sin
        //   y1 = x0*sin + x1*cos
        let x = x.reshape((b_sz, n_head, seq_len, n_embd / 2, 2))?;
        let x0 = x.narrow(D::Minus1, 0, 1)?;
        let x1 = x.narrow(D::Minus1, 1, 1)?;
        let y0 = (x0.broadcast_mul(&cos)? - x1.broadcast_mul(&sin)?)?;
        let y1 = (x0.broadcast_mul(&sin)? + x1.broadcast_mul(&cos)?)?;
        let rope = Tensor::cat(&[y0, y1], D::Minus1)?;
        let rope = rope.flatten_from(D::Minus2)?;
        Ok(rope)
    }

    fn forward_attn(&mut self, x: &Tensor, mask: &Tensor, index_pos: usize) -> Result<Tensor> {
        let _enter = self.span_attn.enter();
        let (b_sz, seq_len, n_embd) = x.dims3()?;
        let q = self.attention_wq.forward(x)?;
        let k = self.attention_wk.forward(x)?;
        let v = self.attention_wv.forward(x)?;

        let q = q
            .reshape((b_sz, seq_len, self.n_head, self.head_dim))?
            .transpose(1, 2)?;
        let k = k
            .reshape((b_sz, seq_len, self.n_kv_head, self.head_dim))?
            .transpose(1, 2)?;
        let v = v
            .reshape((b_sz, seq_len, self.n_kv_head, self.head_dim))?
            .transpose(1, 2)?;

        let q = self.apply_rotary_emb(&q, index_pos)?;
        let k = self.apply_rotary_emb(&k, index_pos)?;

        let (k, v) = match &self.kv_cache {
            None => (k, v),
            Some((k_cache, v_cache)) => {
                let k = Tensor::cat(&[k_cache, &k], 2)?.contiguous()?;
                let v = Tensor::cat(&[v_cache, &v], 2)?.contiguous()?;
                (k, v)
            }
        };
        self.kv_cache = Some((k.clone(), v.clone()));

        // Support for MQA, useful for 70B models.
        let k = self.repeat_kv(k)?;
        let v = self.repeat_kv(v)?;

        let att = (q.matmul(&k.t()?)? / (self.head_dim as f64).sqrt())?;
        let mask = mask.broadcast_as(att.shape())?;
        let att = masked_fill(&att, &mask, f32::NEG_INFINITY)?;
        let att = candle_nn::ops::softmax(&att, D::Minus1)?;
        // Convert to contiguous as matmul doesn't support strided vs for now.
        let y = att.matmul(&v.contiguous()?)?;
        let y = y.transpose(1, 2)?.reshape(&[b_sz, seq_len, n_embd])?;
        let y = self.attention_wo.forward(&y)?;
        Ok(y)
    }

    fn repeat_kv(&self, x: Tensor) -> Result<Tensor> {
        let n_rep = self.n_head / self.n_kv_head;
        if n_rep == 1 {
            Ok(x)
        } else {
            let (b_sz, n_kv_head, seq_len, head_dim) = x.dims4()?;
            let x = x
                .unsqueeze(2)?
                .expand((b_sz, n_kv_head, n_rep, seq_len, head_dim))?
                .reshape((b_sz, n_kv_head * n_rep, seq_len, head_dim))?;
            Ok(x)
        }
    }
}

struct ModelWeights {
    tok_embeddings: Embedding,
    layers: Vec<LayerWeights>,
    norm: RmsNorm,
    output: QMatMul,
    masks: HashMap<usize, Tensor>,
    span: tracing::Span,
    span_output: tracing::Span,
}

struct WeightMap(HashMap<String, QTensor>);
impl WeightMap {
    fn get(&mut self, name: &str) -> Result<QTensor> {
        match self.0.remove(name) {
            None => candle::bail!("cannot find tensor with name '{name}'"),
            Some(tensor) => Ok(tensor),
        }
    }
}

impl ModelWeights {
    fn new(mut ct: Content, gqa: usize) -> Result<Self> {
        let cpu = &Device::Cpu;
        let head_dim = (ct.hparams.n_embd / ct.hparams.n_head) as usize;

        // precompute freqs_cis
        let theta: Vec<_> = (0..head_dim)
            .step_by(2)
            .map(|i| 1f32 / 10000f32.powf(i as f32 / head_dim as f32))
            .collect();
        let theta = Tensor::new(theta.as_slice(), &Device::Cpu)?;
        let idx_theta = Tensor::arange(0, MAX_SEQ_LEN as u32, &Device::Cpu)?
            .to_dtype(DType::F32)?
            .reshape((MAX_SEQ_LEN, 1))?
            .matmul(&theta.reshape((1, theta.elem_count()))?)?;
        let cos = idx_theta.cos()?;
        let sin = idx_theta.sin()?;

        let tok_embeddings = ct.remove("tok_embeddings.weight")?;
        let tok_embeddings = tok_embeddings.dequantize(cpu)?;
        let norm = RmsNorm::new(ct.remove("norm.weight")?)?;
        let output = ct.remove("output.weight")?;
        let mut layers = Vec::with_capacity(ct.hparams.n_layer as usize);
        for layer_idx in 0..ct.hparams.n_layer {
            let prefix = format!("layers.{layer_idx}");
            let attention_wq = ct.remove(&format!("layers.{layer_idx}.attention.wq.weight"))?;
            let attention_wk = ct.remove(&format!("{prefix}.attention.wk.weight"))?;
            let attention_wv = ct.remove(&format!("{prefix}.attention.wv.weight"))?;
            let attention_wo = ct.remove(&format!("{prefix}.attention.wo.weight"))?;
            let feed_forward_w1 = ct.remove(&format!("{prefix}.feed_forward.w1.weight"))?;
            let feed_forward_w2 = ct.remove(&format!("{prefix}.feed_forward.w2.weight"))?;
            let feed_forward_w3 = ct.remove(&format!("{prefix}.feed_forward.w3.weight"))?;
            let attention_norm = ct.remove(&format!("{prefix}.attention_norm.weight"))?;
            let ffn_norm = ct.remove(&format!("{prefix}.ffn_norm.weight"))?;
            let span_attn = tracing::span!(tracing::Level::TRACE, "attn");
            let span_rot = tracing::span!(tracing::Level::TRACE, "attn-rot");
            let span_mlp = tracing::span!(tracing::Level::TRACE, "attn-mlp");
            layers.push(LayerWeights {
                attention_wq: QMatMul::from_qtensor(attention_wq),
                attention_wk: QMatMul::from_qtensor(attention_wk),
                attention_wv: QMatMul::from_qtensor(attention_wv),
                attention_wo: QMatMul::from_qtensor(attention_wo),
                attention_norm: RmsNorm::new(attention_norm)?,
                feed_forward_w1: QMatMul::from_qtensor(feed_forward_w1),
                feed_forward_w2: QMatMul::from_qtensor(feed_forward_w2),
                feed_forward_w3: QMatMul::from_qtensor(feed_forward_w3),
                ffn_norm: RmsNorm::new(ffn_norm)?,
                n_head: ct.hparams.n_head as usize,
                n_kv_head: ct.hparams.n_head as usize / gqa,
                head_dim: (ct.hparams.n_embd / ct.hparams.n_head) as usize,
                cos: cos.clone(),
                sin: sin.clone(),
                kv_cache: None,
                span_attn,
                span_rot,
                span_mlp,
            })
        }
        let span = tracing::span!(tracing::Level::TRACE, "model");
        let span_output = tracing::span!(tracing::Level::TRACE, "output");
        Ok(Self {
            tok_embeddings: Embedding::new(tok_embeddings, ct.hparams.n_embd as usize),
            layers,
            norm,
            output: QMatMul::from_qtensor(output),
            masks: HashMap::new(),
            span,
            span_output,
        })
    }

    fn mask(&mut self, t: usize) -> Result<Tensor> {
        if let Some(mask) = self.masks.get(&t) {
            Ok(mask.clone())
        } else {
            let mask: Vec<_> = (0..t)
                .flat_map(|i| (0..t).map(move |j| u8::from(j > i)))
                .collect();
            let mask = Tensor::from_slice(&mask, (t, t), &Device::Cpu)?;
            self.masks.insert(t, mask.clone());
            Ok(mask)
        }
    }

    fn forward(&mut self, x: &Tensor, index_pos: usize) -> Result<Tensor> {
        let (_b_sz, seq_len) = x.dims2()?;
        let mask = self.mask(seq_len)?;
        let _enter = self.span.enter();
        let mut layer_in = self.tok_embeddings.forward(x)?;
        for layer in self.layers.iter_mut() {
            let x = layer_in;
            let residual = &x;
            let x = layer.attention_norm.forward(&x)?;
            let attn = layer.forward_attn(&x, &mask, index_pos)?;
            let x = (attn + residual)?;

            // MLP
            let _enter = layer.span_mlp.enter();
            let residual = &x;
            let x = layer.ffn_norm.forward(&x)?;
            let w1 = layer.feed_forward_w1.forward(&x)?;
            let w3 = layer.feed_forward_w3.forward(&x)?;
            let mlp = layer
                .feed_forward_w2
                .forward(&(candle_nn::ops::silu(&w1)? * w3)?)?;
            layer_in = (mlp + residual)?;
        }
        let x = self.norm.forward(&layer_in)?;
        let x = x.i((.., seq_len - 1, ..))?;
        let _enter = self.span_output.enter();
        self.output.forward(&x)
    }
}

#[derive(Clone, Debug, Copy, ValueEnum)]
enum Which {
    #[value(name = "7b")]
    L7b,
    #[value(name = "13b")]
    L13b,
    #[value(name = "70b")]
    L70b,
}

#[derive(Parser, Debug)]
#[command(author, version, about, long_about = None)]
struct Args {
    /// GGML file to load, typically a .bin file generated by the quantize command from llama.cpp
    #[arg(long)]
    model: Option<String>,

    /// The initial prompt.
    #[arg(long)]
    prompt: Option<String>,

    /// The length of the sample to generate (in tokens).
    #[arg(short = 'n', long, default_value_t = 100)]
    sample_len: usize,

    /// The tokenizer config in json format.
    #[arg(long)]
    tokenizer: Option<String>,

    /// The temperature used to generate samples.
    #[arg(long)]
    temperature: Option<f64>,

    /// The seed to use when generating random samples.
    #[arg(long, default_value_t = 299792458)]
    seed: u64,

    /// Enable tracing (generates a trace-timestamp.json file).
    #[arg(long)]
    tracing: bool,

    /// Display the token for the specified prompt.
    #[arg(long)]
    verbose_prompt: bool,

    /// Penalty to be applied for repeating tokens, 1. means no penalty.
    #[arg(long, default_value_t = 1.0)]
    repeat_penalty: f32,

    /// The context size to consider for the repeat penalty.
    #[arg(long, default_value_t = 64)]
    repeat_last_n: usize,

    /// The model size to use.
    #[arg(long, default_value = "7b")]
    which: Which,

    /// Group-Query Attention, use 8 for the 70B version of LLaMAv2.
    #[arg(long)]
    gqa: Option<usize>,
}

impl Args {
    fn tokenizer(&self) -> anyhow::Result<Tokenizer> {
        let tokenizer_path = match &self.tokenizer {
            Some(config) => std::path::PathBuf::from(config),
            None => {
                let api = hf_hub::api::sync::Api::new()?;
                let api = api.model("hf-internal-testing/llama-tokenizer".to_string());
                api.get("tokenizer.json")?
            }
        };
        Tokenizer::from_file(tokenizer_path).map_err(anyhow::Error::msg)
    }

    fn model(&self) -> anyhow::Result<std::path::PathBuf> {
        let model_path = match &self.model {
            Some(config) => std::path::PathBuf::from(config),
            None => {
                let (repo, filename) = match self.which {
                    Which::L7b => ("TheBloke/Llama-2-7B-GGML", "llama-2-7b.ggmlv3.q4_0.bin"),
                    Which::L13b => ("TheBloke/Llama-2-13B-GGML", "llama-2-13b.ggmlv3.q4_0.bin"),
                    Which::L70b => ("TheBloke/Llama-2-70B-GGML", "llama-2-70b.ggmlv3.q4_0.bin"),
                };
                let api = hf_hub::api::sync::Api::new()?;
                let api = api.model(repo.to_string());
                api.get(filename)?
            }
        };
        Ok(model_path)
    }
}

fn print_token(next_token: u32, tokenizer: &Tokenizer) {
    // Extracting the last token as a string is complicated, here we just apply some simple
    // heuristics as it seems to work well enough for this example. See the following for more
    // details:
    // https://github.com/huggingface/tokenizers/issues/1141#issuecomment-1562644141
    if let Some(text) = tokenizer.id_to_token(next_token) {
        let text = text.replace('▁', " ");
        let ascii = text
            .strip_prefix("<0x")
            .and_then(|t| t.strip_suffix('>'))
            .and_then(|t| u8::from_str_radix(t, 16).ok());
        match ascii {
            None => print!("{text}"),
            Some(ascii) => {
                if let Some(chr) = char::from_u32(ascii as u32) {
                    if chr.is_ascii() {
                        print!("{chr}")
                    }
                }
            }
        }
        let _ = std::io::stdout().flush();
    }
}

fn apply_repeat_penalty(logits: &Tensor, penalty: f32, context: &[u32]) -> Result<Tensor> {
    let mut logits = logits.to_vec1::<f32>()?;
    let context: std::collections::HashSet<_> = context.iter().collect();
    for (token_id, logit) in logits.iter_mut().enumerate() {
        if context.contains(&(token_id as u32)) {
            if *logit >= 0. {
                *logit /= penalty
            } else {
                *logit *= penalty
            }
        }
    }
    let logits_len = logits.len();
    Tensor::from_vec(logits, logits_len, &Device::Cpu)
}

fn main() -> anyhow::Result<()> {
    use tracing_chrome::ChromeLayerBuilder;
    use tracing_subscriber::prelude::*;

    let args = Args::parse();
    let _guard = if args.tracing {
        let (chrome_layer, guard) = ChromeLayerBuilder::new().build();
        tracing_subscriber::registry().with(chrome_layer).init();
        Some(guard)
    } else {
        None
    };

    println!(
        "avx: {}, neon: {}, simd128: {}, f16c: {}",
        candle::utils::with_avx(),
        candle::utils::with_neon(),
        candle::utils::with_simd128(),
        candle::utils::with_f16c()
    );

    let mut file = std::fs::File::open(&args.model()?)?;
    let start = std::time::Instant::now();
    let model = Content::read(&mut file)?;

    let mut total_size_in_bytes = 0;
    for (_, tensor) in model.tensors.iter() {
        let elem_count = tensor.shape().elem_count();
        total_size_in_bytes += elem_count * tensor.dtype().type_size() / tensor.dtype().blck_size();
    }
    let total_size = if total_size_in_bytes < 1_000 {
        format!("{}B", total_size_in_bytes)
    } else if total_size_in_bytes < 1_000_000 {
        format!("{:.2}KB", total_size_in_bytes as f64 / 1e3)
    } else if total_size_in_bytes < 1_000_000_000 {
        format!("{:.2}MB", total_size_in_bytes as f64 / 1e6)
    } else {
        format!("{:.2}GB", total_size_in_bytes as f64 / 1e9)
    };

    println!(
        "loaded {:?} tensors ({}) in {:.2}s",
        model.tensors.len(),
        total_size,
        start.elapsed().as_secs_f32(),
    );
    println!("params: {:?}", model.hparams);
    let default_gqa = match args.which {
        Which::L7b | Which::L13b => 1,
        Which::L70b => 8,
    };
    let mut model = ModelWeights::new(model, args.gqa.unwrap_or(default_gqa))?;
    println!("model built");

    let tokenizer = args.tokenizer()?;
    let prompt = args.prompt.as_ref().map_or(DEFAULT_PROMPT, |p| p.as_str());
    let tokens = tokenizer.encode(prompt, true).map_err(anyhow::Error::msg)?;
    if args.verbose_prompt {
        for (token, id) in tokens.get_tokens().iter().zip(tokens.get_ids().iter()) {
            let token = token.replace('▁', " ").replace("<0x0A>", "\n");
            println!("{id:7} -> '{token}'");
        }
    }

    let prompt_tokens = tokens.get_ids().to_vec();
    let mut all_tokens = vec![];
    let mut logits_processor = LogitsProcessor::new(args.seed, args.temperature);

    print!("{prompt}");

    let start_prompt_processing = std::time::Instant::now();
    let mut next_token = {
        let input = Tensor::new(prompt_tokens.as_slice(), &Device::Cpu)?.unsqueeze(0)?;
        let logits = model.forward(&input, 0)?;
        let logits = logits.squeeze(0)?;
        logits_processor.sample(&logits)?
    };
    let prompt_dt = start_prompt_processing.elapsed();
    all_tokens.push(next_token);
    print_token(next_token, &tokenizer);

    let to_sample = args.sample_len.saturating_sub(1);
    let start_post_prompt = std::time::Instant::now();
    for index in 0..to_sample {
        let input = Tensor::new(&[next_token], &Device::Cpu)?.unsqueeze(0)?;
        let logits = model.forward(&input, prompt_tokens.len() + index)?;
        let logits = logits.squeeze(0)?;
        let logits = if args.repeat_penalty == 1. {
            logits
        } else {
            let start_at = all_tokens.len().saturating_sub(args.repeat_last_n);
            apply_repeat_penalty(&logits, args.repeat_penalty, &all_tokens[start_at..])?
        };
        next_token = logits_processor.sample(&logits)?;
        all_tokens.push(next_token);
        print_token(next_token, &tokenizer);
    }
    let dt = start_post_prompt.elapsed();
    println!(
        "\n\n{:4} prompt tokens processed: {:.2} token/s",
        prompt_tokens.len(),
        prompt_tokens.len() as f64 / prompt_dt.as_secs_f64(),
    );
    println!(
        "{:4} tokens generated: {:.2} token/s",
        to_sample,
        to_sample as f64 / dt.as_secs_f64(),
    );
    Ok(())
}