Skip to content

More vision models with TP #1200

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Mar 13, 2025
Merged

More vision models with TP #1200

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b5d7593
More models for tp
EricLBuehler Mar 12, 2025
456ba49
Fix clippy
EricLBuehler Mar 13, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions mistralrs-core/src/vision_models/mllama/vision.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -260,15 +260,15 @@ impl MLlamaMlp {
cfg.hidden_size,
cfg.intermediate_size,
&None,
false,
true,
comm,
vb.pp("fc1"),
)?,
fc2: RowParallelLayer::new(
cfg.intermediate_size,
cfg.hidden_size,
&None,
false,
true,
comm,
vb.pp("fc2"),
)?,
Expand Down
4 changes: 2 additions & 2 deletions mistralrs-core/src/vision_models/qwen2_5_vl/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,8 +50,8 @@ impl Qwen2_5VLModel {
let vision = Qwen2_5VLVisionModel::new(
&cfg.vision_config,
vb.pp("visual")
.set_device(normal_loading_metadata.real_device.clone())
.set_dtype(DType::F32),
.set_device(normal_loading_metadata.real_device.clone()),
&normal_loading_metadata.mapper.get_comm_for(0)?,
)?;
let text = Qwen2_5VLTextModel::new(
cfg,
Expand Down
68 changes: 52 additions & 16 deletions mistralrs-core/src/vision_models/qwen2_5_vl/vision.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@ use std::sync::Arc;

use candle_core::{DType, Device, IndexOp, Result, Tensor, D};
use candle_nn::{Linear, Module};
use mistralrs_quant::{QuantMethod, ShardedVarBuilder};
use mistralrs_quant::{ColumnParallelLayer, QuantMethod, RowParallelLayer, ShardedVarBuilder};

use crate::{
layers::{self, Activation, Conv3dConfig, Conv3dNoBias, MatMul, RmsNorm},
Expand Down Expand Up @@ -64,11 +64,38 @@ struct VisionMlp {
}

impl VisionMlp {
fn new(dim: usize, hidden_dim: usize, act: Activation, vb: ShardedVarBuilder) -> Result<Self> {
fn new(
dim: usize,
hidden_dim: usize,
act: Activation,
vb: ShardedVarBuilder,
comm: &Arc<mistralrs_quant::Comm>,
) -> Result<Self> {
Ok(Self {
gate_proj: mistralrs_quant::linear(dim, hidden_dim, &None, vb.pp("gate_proj"))?,
up_proj: mistralrs_quant::linear(dim, hidden_dim, &None, vb.pp("up_proj"))?,
down_proj: mistralrs_quant::linear(hidden_dim, dim, &None, vb.pp("down_proj"))?,
gate_proj: ColumnParallelLayer::new(
dim,
hidden_dim,
&None,
true,
comm,
vb.pp("gate_proj"),
)?,
up_proj: ColumnParallelLayer::new(
dim,
hidden_dim,
&None,
true,
comm,
vb.pp("up_proj"),
)?,
down_proj: RowParallelLayer::new(
hidden_dim,
dim,
&None,
true,
comm,
vb.pp("down_proj"),
)?,
act,
})
}
Expand Down Expand Up @@ -102,11 +129,10 @@ fn rotate_half(xs: &Tensor) -> Result<Tensor> {
}

fn apply_rotary_pos_emb_vision(xs: &Tensor, freqs: &Tensor) -> Result<Tensor> {
let xs = xs.to_dtype(DType::F32)?;
let cos = freqs.cos()?.unsqueeze(D::Minus2)?;
let sin = freqs.sin()?.unsqueeze(D::Minus2)?;
let cos = freqs.cos()?.unsqueeze(D::Minus2)?.to_dtype(xs.dtype())?;
let sin = freqs.sin()?.unsqueeze(D::Minus2)?.to_dtype(xs.dtype())?;

xs.broadcast_mul(&cos)? + rotate_half(&xs)?.broadcast_mul(&sin)
xs.broadcast_mul(&cos)? + rotate_half(xs)?.broadcast_mul(&sin)
}

// https://github.com/huggingface/transformers/blob/a769ed45e17c44fd17b85c025863c4e4f2f73634/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py#L325
Expand Down Expand Up @@ -182,7 +208,11 @@ struct VisionBlock {
}

impl VisionBlock {
fn new(cfg: &VisionConfig, vb: ShardedVarBuilder) -> Result<Self> {
fn new(
cfg: &VisionConfig,
vb: ShardedVarBuilder,
comm: &Arc<mistralrs_quant::Comm>,
) -> Result<Self> {
let norm1 = RmsNorm::new(cfg.hidden_size, 1e-6, vb.pp("norm1"))?;
let norm2 = RmsNorm::new(cfg.hidden_size, 1e-6, vb.pp("norm2"))?;

Expand All @@ -191,6 +221,7 @@ impl VisionBlock {
cfg.intermediate_size,
cfg.hidden_act,
vb.pp("mlp"),
comm,
)?;
let attn = VisionAttention::new(cfg.hidden_size, cfg.num_heads, vb.pp("attn"))?;

Expand Down Expand Up @@ -290,10 +321,14 @@ pub struct Qwen2_5VLVisionModel {
}

impl Qwen2_5VLVisionModel {
pub fn new(cfg: &VisionConfig, vb: ShardedVarBuilder) -> Result<Self> {
pub fn new(
cfg: &VisionConfig,
vb: ShardedVarBuilder,
comm: &Arc<mistralrs_quant::Comm>,
) -> Result<Self> {
let mut blocks = Vec::new();
for i in 0..cfg.depth {
blocks.push(VisionBlock::new(cfg, vb.pp(format!("blocks.{i}")))?);
blocks.push(VisionBlock::new(cfg, vb.pp(format!("blocks.{i}")), comm)?);
}

let patch_merger = PatchMerger::new(
Expand Down Expand Up @@ -454,6 +489,7 @@ impl Qwen2_5VLVisionModel {
rotary_pos_emb = rotary_pos_emb.index_select(&window_index, 0)?;
rotary_pos_emb = rotary_pos_emb.reshape((seq_len, ()))?;
rotary_pos_emb = Tensor::cat(&[&rotary_pos_emb; 2], D::Minus1)?;
rotary_pos_emb = rotary_pos_emb.to_dtype(xs.dtype())?;

let grid_thw = grid_thw.to_device(&Device::Cpu)?;
let cu_seqlens = (grid_thw.i((.., 1))? * grid_thw.i((.., 2))?)?
Expand All @@ -470,13 +506,13 @@ impl Qwen2_5VLVisionModel {
cu_seqlens => {
let mut attention_mask =
Tensor::full(f32::MIN, (1, seq_len, seq_len), xs.device())?
.to_dtype(DType::F32)?;
.to_dtype(xs.dtype())?;
for i in 1..cu_seqlens.len() {
let a = cu_seqlens[i - 1] as usize;
let b = cu_seqlens[i] as usize;
attention_mask = attention_mask.slice_assign(
&[&.., &(a..b), &(a..b)],
&Tensor::zeros((1, b - a, b - a), DType::F32, xs.device())?,
&Tensor::zeros((1, b - a, b - a), xs.dtype(), xs.device())?,
)?;
}
Some(attention_mask)
Expand All @@ -487,13 +523,13 @@ impl Qwen2_5VLVisionModel {
cu_seqlens => {
let mut attention_mask =
Tensor::full(f32::MIN, (1, seq_len, seq_len), xs.device())?
.to_dtype(DType::F32)?;
.to_dtype(xs.dtype())?;
for i in 1..cu_seqlens.len() {
let a = cu_seqlens[i - 1] as usize;
let b = cu_seqlens[i] as usize;
attention_mask = attention_mask.slice_assign(
&[&.., &(a..b), &(a..b)],
&Tensor::zeros((1, b - a, b - a), DType::F32, xs.device())?,
&Tensor::zeros((1, b - a, b - a), xs.dtype(), xs.device())?,
)?;
}
Some(attention_mask)
Expand Down
4 changes: 2 additions & 2 deletions mistralrs-core/src/vision_models/qwen2vl/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,8 +50,8 @@ impl Qwen2VLModel {
let vision = Qwen2VLVisionModel::new(
&cfg.vision_config,
vb.pp("visual")
.set_device(normal_loading_metadata.real_device.clone())
.set_dtype(DType::F32),
.set_device(normal_loading_metadata.real_device.clone()),
&normal_loading_metadata.mapper.get_comm_for(0)?,
)?;
let text = Qwen2VLTextModel::new(
cfg,
Expand Down
45 changes: 32 additions & 13 deletions mistralrs-core/src/vision_models/qwen2vl/vision.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@ use std::sync::Arc;

use candle_core::{DType, Device, IndexOp, Result, Tensor, D};
use candle_nn::{LayerNorm, Linear, Module};
use mistralrs_quant::{QuantMethod, ShardedVarBuilder};
use mistralrs_quant::{ColumnParallelLayer, QuantMethod, ShardedVarBuilder};

use crate::{
layers::{self, layer_norm, Activation, Conv3dConfig, Conv3dNoBias, MatMul},
Expand Down Expand Up @@ -63,10 +63,16 @@ struct VisionMlp {
}

impl VisionMlp {
fn new(dim: usize, hidden_dim: usize, act: Activation, vb: ShardedVarBuilder) -> Result<Self> {
fn new(
dim: usize,
hidden_dim: usize,
act: Activation,
vb: ShardedVarBuilder,
comm: &Arc<mistralrs_quant::Comm>,
) -> Result<Self> {
Ok(Self {
fc1: mistralrs_quant::linear(dim, hidden_dim, &None, vb.pp("fc1"))?,
fc2: mistralrs_quant::linear(hidden_dim, dim, &None, vb.pp("fc2"))?,
fc1: ColumnParallelLayer::new(dim, hidden_dim, &None, true, comm, vb.pp("fc1"))?,
fc2: ColumnParallelLayer::new(hidden_dim, dim, &None, true, comm, vb.pp("fc2"))?,
act,
})
}
Expand All @@ -85,11 +91,10 @@ fn rotate_half(xs: &Tensor) -> Result<Tensor> {
}

fn apply_rotary_pos_emb_vision(xs: &Tensor, freqs: &Tensor) -> Result<Tensor> {
let xs = xs.to_dtype(DType::F32)?;
let cos = freqs.cos()?;
let sin = freqs.sin()?;

xs.broadcast_mul(&cos)? + rotate_half(&xs)?.broadcast_mul(&sin)
xs.broadcast_mul(&cos)? + rotate_half(xs)?.broadcast_mul(&sin)
}

// https://github.com/huggingface/transformers/blob/a769ed45e17c44fd17b85c025863c4e4f2f73634/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py#L325
Expand Down Expand Up @@ -165,12 +170,22 @@ struct VisionBlock {
}

impl VisionBlock {
fn new(cfg: &VisionConfig, vb: ShardedVarBuilder) -> Result<Self> {
fn new(
cfg: &VisionConfig,
vb: ShardedVarBuilder,
comm: &Arc<mistralrs_quant::Comm>,
) -> Result<Self> {
let norm1 = layer_norm(cfg.embed_dim, 1e-6, vb.pp("norm1"))?;
let norm2 = layer_norm(cfg.embed_dim, 1e-6, vb.pp("norm2"))?;

let mlp_hidden_dim = (cfg.embed_dim as f64 * cfg.mlp_ratio) as usize;
let mlp = VisionMlp::new(cfg.embed_dim, mlp_hidden_dim, cfg.hidden_act, vb.pp("mlp"))?;
let mlp = VisionMlp::new(
cfg.embed_dim,
mlp_hidden_dim,
cfg.hidden_act,
vb.pp("mlp"),
comm,
)?;
let attn = VisionAttention::new(cfg.embed_dim, cfg.num_heads, vb.pp("attn"))?;

Ok(Self {
Expand Down Expand Up @@ -265,10 +280,14 @@ pub struct Qwen2VLVisionModel {
}

impl Qwen2VLVisionModel {
pub fn new(cfg: &VisionConfig, vb: ShardedVarBuilder) -> Result<Self> {
pub fn new(
cfg: &VisionConfig,
vb: ShardedVarBuilder,
comm: &Arc<mistralrs_quant::Comm>,
) -> Result<Self> {
let mut blocks = Vec::new();
for i in 0..cfg.depth {
blocks.push(VisionBlock::new(cfg, vb.pp(format!("blocks.{i}")))?);
blocks.push(VisionBlock::new(cfg, vb.pp(format!("blocks.{i}")), comm)?);
}

let patch_merger = PatchMerger::new(
Expand Down Expand Up @@ -342,7 +361,7 @@ impl Qwen2VLVisionModel {
.unsqueeze(1)?
.repeat((1, 1, 2))?
.unsqueeze(0)?
.to_dtype(DType::F32)?;
.to_dtype(xs.dtype())?;

let grid_thw = grid_thw.to_device(&Device::Cpu)?;
let cu_seqlens = (grid_thw.i((.., 1))? * grid_thw.i((.., 2))?)?
Expand All @@ -359,13 +378,13 @@ impl Qwen2VLVisionModel {
cu_seqlens => {
let mut attention_mask =
Tensor::full(f32::MIN, (1, seq_len, seq_len), xs.device())?
.to_dtype(DType::F32)?;
.to_dtype(xs.dtype())?;
for i in 1..cu_seqlens.len() {
let a = cu_seqlens[i - 1] as usize;
let b = cu_seqlens[i] as usize;
attention_mask = attention_mask.slice_assign(
&[&.., &(a..b), &(a..b)],
&Tensor::zeros((1, b - a, b - a), DType::F32, xs.device())?,
&Tensor::zeros((1, b - a, b - a), xs.dtype(), xs.device())?,
)?;
}
Some(attention_mask)
Expand Down
Loading