Compare commits
10 Commits
82005f8eeb
...
relaxed-re
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
791c38dcc3 | ||
|
|
0af78a9882 | ||
|
|
fa5efbf235 | ||
|
|
59a7ac427d | ||
|
|
e3393042e5 | ||
|
|
08a4e8a7fb | ||
|
|
b582d340b0 | ||
|
|
474ba1a1b8 | ||
|
|
de771fcb05 | ||
|
|
f32d429db5 |
@@ -4,12 +4,8 @@ Axolotl Plugin for Relaxed Recursive Transformers
|
||||
|
||||
import logging
|
||||
|
||||
from transformers import AutoConfig, AutoModel, AutoModelForCausalLM
|
||||
|
||||
from axolotl.integrations.base import BasePlugin
|
||||
from axolotl.integrations.rrt.modeling import register_rrt_model
|
||||
from axolotl.integrations.rrt.modeling.modeling_rrt_llama import RelaxedRecursiveLlamaConfig, \
|
||||
RelaxedRecursiveLlamaModel, RelaxedRecursiveLlamaForCausalLM
|
||||
|
||||
LOG = logging.getLogger(__name__)
|
||||
|
||||
@@ -20,23 +16,10 @@ class RelaxedRecursiveTransformerPlugin(BasePlugin):
|
||||
"""
|
||||
|
||||
def get_input_args(self):
|
||||
return "axolotl.integrations.rrt.RelaxedRecursiveTransformerArgs"
|
||||
return "axolotl.integrations.rrt.args.RelaxedRecursiveTransformerArgs"
|
||||
|
||||
def register(self):
|
||||
LOG.info(
|
||||
"Registering Relaxed Recursive Transformers modeling with transformers"
|
||||
)
|
||||
register_rrt_model()
|
||||
|
||||
|
||||
def register_rrt_model():
|
||||
"""
|
||||
Register Relaxed Recursive Transformers model with transformers
|
||||
"""
|
||||
|
||||
# Register configs
|
||||
AutoConfig.register("llama-rrt", RelaxedRecursiveLlamaConfig)
|
||||
|
||||
# Register models
|
||||
AutoModel.register("llama-rrt", RelaxedRecursiveLlamaConfig, RelaxedRecursiveLlamaModel)
|
||||
AutoModelForCausalLM.register("llama-rrt", RelaxedRecursiveLlamaConfig, RelaxedRecursiveLlamaForCausalLM)
|
||||
|
||||
@@ -1,3 +1,7 @@
|
||||
"""
|
||||
Axolotl config args for Relaxed Recursive Transformers plugin
|
||||
"""
|
||||
|
||||
from pydantic import BaseModel
|
||||
|
||||
|
||||
@@ -5,4 +9,3 @@ class RelaxedRecursiveTransformerArgs(BaseModel):
|
||||
"""
|
||||
Arguments pertaining to the Relaxed Recursive Transformer model.
|
||||
"""
|
||||
...
|
||||
|
||||
@@ -1,4 +1,8 @@
|
||||
"""
|
||||
cli script for converting a pretrained model to a relaxed recursive transformer model
|
||||
"""
|
||||
import json
|
||||
import logging
|
||||
import math
|
||||
import os
|
||||
import re
|
||||
@@ -10,15 +14,19 @@ import torch
|
||||
from huggingface_hub import snapshot_download, split_torch_state_dict_into_shards
|
||||
from safetensors.torch import save_file
|
||||
from tqdm import tqdm
|
||||
from transformers import AutoConfig
|
||||
from transformers.utils import SAFE_WEIGHTS_NAME, SAFE_WEIGHTS_INDEX_NAME
|
||||
from transformers import AutoConfig, AutoTokenizer
|
||||
from transformers.utils import SAFE_WEIGHTS_INDEX_NAME, SAFE_WEIGHTS_NAME
|
||||
|
||||
from axolotl.integrations.rrt.modeling.modeling_rrt_llama import RelaxedRecursiveLlamaConfig
|
||||
from axolotl.integrations.rrt.modeling.modeling_rrt_llama import (
|
||||
RelaxedRecursiveLlamaConfig,
|
||||
)
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
def extract_layer_number(key):
|
||||
"""Extract layer number from parameter key."""
|
||||
match = re.search(r'layers\.(\d+)\.', key)
|
||||
match = re.search(r"layers\.(\d+)\.", key)
|
||||
return int(match.group(1)) if match else None
|
||||
|
||||
|
||||
@@ -30,28 +38,30 @@ def iter_parameter_weights(model_path, device="mps"):
|
||||
:param device: Computing device
|
||||
:return: generator yielding (parameter key, parameter weight, layer index) tuples
|
||||
"""
|
||||
shards = list(model_path.glob('model*.safetensors'))
|
||||
shards = list(model_path.glob("model*.safetensors"))
|
||||
if not shards:
|
||||
raise ValueError(f"No model shards found in {model_path}")
|
||||
|
||||
for shard in tqdm(shards, desc="Processing shards"):
|
||||
with safetensors.safe_open(shard, framework='pt', device=device) as f:
|
||||
for key in f.keys():
|
||||
layer_idx = extract_layer_number(key)
|
||||
weight = f.get_tensor(key)
|
||||
yield key, weight, layer_idx
|
||||
with safetensors.safe_open(shard, framework="pt", device=device) as f:
|
||||
for key in f.keys():
|
||||
layer_idx = extract_layer_number(key)
|
||||
weight = f.get_tensor(key)
|
||||
yield key, weight, layer_idx
|
||||
|
||||
def iter_recursive_parameter_weights(model_path, modules_to_recurse: list[str], device="mps", recurse_layers=12):
|
||||
|
||||
def iter_recursive_parameter_weights(
|
||||
model_path, modules_to_recurse: list[str], device="mps", recurse_layers=12
|
||||
):
|
||||
# setup placeholder state_dict for recursive weights, need to keep in float32 precision
|
||||
# to avoid precision loss when averaging weights across layers
|
||||
rrt_avg_model_state_dict = {}
|
||||
rrt_avg_model_state_dict: dict[str, list[torch.Tensor]] = {}
|
||||
|
||||
# iterate over all parameter weights in the model shards
|
||||
for key, weight, layer_idx in iter_parameter_weights(model_path, device=device):
|
||||
# get the matching module name in modules_to_recurse for the current parameter key
|
||||
matched_module_name = next(
|
||||
(module for module in modules_to_recurse if module in key),
|
||||
None
|
||||
(module for module in modules_to_recurse if module in key), None
|
||||
)
|
||||
if matched_module_name is None:
|
||||
continue
|
||||
@@ -62,7 +72,9 @@ def iter_recursive_parameter_weights(model_path, modules_to_recurse: list[str],
|
||||
# setup as storage for suffix with torch.stack
|
||||
rrt_avg_model_state_dict[suffix] = [weight.to(torch.float32).detach().cpu()]
|
||||
else:
|
||||
rrt_avg_model_state_dict[suffix].append(weight.to(torch.float32).detach().cpu())
|
||||
rrt_avg_model_state_dict[suffix].append(
|
||||
weight.to(torch.float32).detach().cpu()
|
||||
)
|
||||
|
||||
for module_name in modules_to_recurse:
|
||||
for recurse_idx in range(recurse_layers):
|
||||
@@ -73,8 +85,9 @@ def iter_recursive_parameter_weights(model_path, modules_to_recurse: list[str],
|
||||
|
||||
# compute the decomposed lora diff from the weight base to the actual weight for each module
|
||||
|
||||
|
||||
def low_rank_decomposition(
|
||||
weight: torch.Tensor, max_rank: int
|
||||
weight: torch.Tensor, max_rank: int
|
||||
) -> Tuple[torch.Tensor, torch.Tensor]:
|
||||
"""
|
||||
Decompose a 2D matrix into low-rank matrices L and R using SVD.
|
||||
@@ -83,18 +96,19 @@ def low_rank_decomposition(
|
||||
:param max_rank: The maximum rank of the decomposition
|
||||
:return: A tuple of tensors (L, R)
|
||||
"""
|
||||
# pylint: disable=invalid-name
|
||||
assert (
|
||||
weight.dim() == 2
|
||||
weight.dim() == 2
|
||||
), f"Only support 2D matrix, but input has {weight.dim()} dimensions."
|
||||
assert (
|
||||
max_rank >= 1
|
||||
max_rank >= 1
|
||||
), f"Maximum rank must be a positive integer, but input max_rank={max_rank}."
|
||||
|
||||
dtype = weight.dtype
|
||||
|
||||
U, S, Vh = torch.linalg.svd(weight.float(), full_matrices=False)
|
||||
|
||||
# Distribute S to both to improve numerical precision.
|
||||
# Distribute S to both to improve numerical precision
|
||||
sqrt_S = torch.sqrt(torch.diag(S[:max_rank]))
|
||||
A = sqrt_S @ Vh[:max_rank, :] # shape: [r, cols]
|
||||
B = U[:, :max_rank] @ sqrt_S # shape: [rows, r]
|
||||
@@ -109,7 +123,7 @@ def get_weight_norm(weight, lora_weight, scaling) -> torch.Tensor:
|
||||
return weight_norm
|
||||
|
||||
|
||||
def decompose_delta_weight(layer_weight, avg_weight, alpha, rank):
|
||||
def decompose_delta_weight(layer_weight, avg_weight, alpha, rank, use_dora=True):
|
||||
"""
|
||||
Decompose the difference in directions (ΔV) via SVD,
|
||||
and return (magnitudes, L, R).
|
||||
@@ -122,36 +136,49 @@ def decompose_delta_weight(layer_weight, avg_weight, alpha, rank):
|
||||
base_weight = avg_weight.to(device)
|
||||
final_weight = layer_weight.to(device)
|
||||
|
||||
delta_first_pass = final_weight - base_weight
|
||||
delta_for_svd = final_weight - base_weight
|
||||
|
||||
delta_for_svd = delta_first_pass / scaling
|
||||
# Low-rank factorization of the delta direction
|
||||
lora_A, lora_B = low_rank_decomposition( # pylint: disable=invalid-name
|
||||
delta_for_svd, rank
|
||||
)
|
||||
|
||||
# 3. Low-rank factorization of the delta direction
|
||||
lora_A, lora_B = low_rank_decomposition(delta_for_svd, rank)
|
||||
if use_dora:
|
||||
lora_weight = lora_B @ lora_A
|
||||
weight_norm = get_weight_norm(
|
||||
base_weight.to(lora_A.device), lora_weight, scaling
|
||||
)
|
||||
return lora_A.cpu(), lora_B.cpu(), weight_norm.cpu()
|
||||
|
||||
lora_weight = lora_B @ lora_A
|
||||
# let's rescale the lora weight to have the same magnitude as the base weight
|
||||
|
||||
weight_norm = get_weight_norm(base_weight.to(lora_A.device), lora_weight, scaling)
|
||||
|
||||
return lora_A.cpu(), lora_B.cpu(), weight_norm.cpu()
|
||||
return lora_A.cpu(), lora_B.cpu(), None
|
||||
|
||||
|
||||
def iter_dora_parameter_weights(model_path, avg_recursive_weights, modules_to_recurse: list[str], alpha, rank, device="mps", recurse_layers=12):
|
||||
rrt_avg_model_state_dict = {}
|
||||
|
||||
def iter_dora_parameter_weights(
|
||||
model_path,
|
||||
avg_recursive_weights,
|
||||
modules_to_recurse: list[str],
|
||||
alpha,
|
||||
rank,
|
||||
device="mps",
|
||||
recurse_layers=12,
|
||||
use_dora=True,
|
||||
):
|
||||
# iterate over all parameter weights in the model shards
|
||||
for key, weight, layer_idx in iter_parameter_weights(model_path, device=device):
|
||||
# get the matching module name in modules_to_recurse for the current parameter key
|
||||
matched_module_name = next(
|
||||
(module for module in modules_to_recurse if module in key),
|
||||
None
|
||||
(module for module in modules_to_recurse if module in key), None
|
||||
)
|
||||
if matched_module_name is None:
|
||||
if "input_layernorm" in key:
|
||||
# map to input_layernorm_list in the recursive layers and account for the layer_idx and loop_idx
|
||||
loop_idx = layer_idx // recurse_layers
|
||||
layer_idx = layer_idx % recurse_layers
|
||||
layernorm_key = f"model.layers.{layer_idx}.input_layernorm_list.{loop_idx}.weight"
|
||||
layernorm_key = (
|
||||
f"model.layers.{layer_idx}.input_layernorm_list.{loop_idx}.weight"
|
||||
)
|
||||
yield layernorm_key, weight
|
||||
elif "post_attention_layernorm" in key:
|
||||
# map to input_layernorm_list in the recursive layers and account for the layer_idx and loop_idx
|
||||
@@ -169,22 +196,35 @@ def iter_dora_parameter_weights(model_path, avg_recursive_weights, modules_to_re
|
||||
suffix = f"{layer_idx}.{matched_module_name}"
|
||||
prefix = f"model.layers.{suffix}.weight_base"
|
||||
avg_weight = avg_recursive_weights[prefix]
|
||||
lora_a_key = f"model.layers.{suffix}.lora_A_list.{loop_idx}"
|
||||
lora_b_key = f"model.layers.{suffix}.lora_B_list.{loop_idx}"
|
||||
lora_magnitude_key = f"model.layers.{suffix}.lora_magnitude_vector_list.{loop_idx}"
|
||||
lora_a, lora_b, lora_magnitude = decompose_delta_weight(weight, avg_weight, alpha, rank)
|
||||
lora_a_key = f"model.layers.{suffix}.lora_A_list.{loop_idx}"
|
||||
lora_b_key = f"model.layers.{suffix}.lora_B_list.{loop_idx}"
|
||||
lora_magnitude_key = (
|
||||
f"model.layers.{suffix}.lora_magnitude_vector_list.{loop_idx}"
|
||||
)
|
||||
lora_a, lora_b, lora_magnitude = decompose_delta_weight(
|
||||
weight,
|
||||
avg_weight,
|
||||
alpha,
|
||||
rank,
|
||||
use_dora=use_dora,
|
||||
)
|
||||
yield lora_a_key, lora_a
|
||||
yield lora_b_key, lora_b
|
||||
yield lora_magnitude_key, lora_magnitude
|
||||
if use_dora:
|
||||
yield lora_magnitude_key, lora_magnitude
|
||||
|
||||
|
||||
def save_state_dict_to_safetensors(state_dict, save_directory):
|
||||
os.makedirs(save_directory, exist_ok=True)
|
||||
weights_name = SAFE_WEIGHTS_NAME
|
||||
|
||||
filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
|
||||
filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(
|
||||
".safetensors", "{suffix}.safetensors"
|
||||
)
|
||||
state_dict_split = split_torch_state_dict_into_shards(
|
||||
state_dict, filename_pattern=filename_pattern, max_shard_size="1GB"
|
||||
)
|
||||
# pylint: disable=duplicate-code
|
||||
# Save index if sharded
|
||||
index = None
|
||||
if state_dict_split.is_sharded:
|
||||
@@ -205,10 +245,10 @@ def save_state_dict_to_safetensors(state_dict, save_directory):
|
||||
reg = re.compile(r"(.*?)-\d{5}-of-\d{5}")
|
||||
|
||||
if (
|
||||
filename.startswith(weights_no_suffix)
|
||||
and os.path.isfile(full_filename)
|
||||
and filename not in state_dict_split.filename_to_tensors.keys()
|
||||
and reg.fullmatch(filename_no_suffix) is not None
|
||||
filename.startswith(weights_no_suffix)
|
||||
and os.path.isfile(full_filename)
|
||||
and filename not in state_dict_split.filename_to_tensors.keys()
|
||||
and reg.fullmatch(filename_no_suffix) is not None
|
||||
):
|
||||
os.remove(full_filename)
|
||||
|
||||
@@ -219,7 +259,9 @@ def save_state_dict_to_safetensors(state_dict, save_directory):
|
||||
shard[tensor] = state_dict[tensor].contiguous()
|
||||
del state_dict[tensor]
|
||||
|
||||
save_file(shard, os.path.join(save_directory, shard_file), metadata={"format": "pt"})
|
||||
save_file(
|
||||
shard, os.path.join(save_directory, shard_file), metadata={"format": "pt"}
|
||||
)
|
||||
|
||||
del state_dict
|
||||
|
||||
@@ -234,7 +276,24 @@ def save_state_dict_to_safetensors(state_dict, save_directory):
|
||||
content = json.dumps(index, indent=2, sort_keys=True) + "\n"
|
||||
f.write(content)
|
||||
|
||||
def convert_llama_to_rrt(model_name, output_dir, recurse_layers: int = 12, rank=32, alpha=32, device="mps"):
|
||||
|
||||
def convert_llama_to_rrt(
|
||||
model_name,
|
||||
output_dir,
|
||||
recurse_layers: int = 12,
|
||||
rank=32,
|
||||
alpha=32,
|
||||
device=None,
|
||||
use_dora=True,
|
||||
):
|
||||
if not device:
|
||||
if torch.backends.mps.is_available():
|
||||
device = "mps"
|
||||
elif torch.cuda.is_available():
|
||||
device = "cuda"
|
||||
else:
|
||||
device = "cpu"
|
||||
|
||||
modules_to_recurse = [
|
||||
"self_attn.q_proj",
|
||||
"self_attn.k_proj",
|
||||
@@ -246,6 +305,7 @@ def convert_llama_to_rrt(model_name, output_dir, recurse_layers: int = 12, rank=
|
||||
]
|
||||
|
||||
config = AutoConfig.from_pretrained(model_name)
|
||||
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
||||
num_hidden_layers = config.num_hidden_layers
|
||||
if num_hidden_layers % recurse_layers != 0:
|
||||
raise ValueError(
|
||||
@@ -253,21 +313,41 @@ def convert_llama_to_rrt(model_name, output_dir, recurse_layers: int = 12, rank=
|
||||
f"divisible by the recurse layers ({recurse_layers})"
|
||||
)
|
||||
|
||||
config = RelaxedRecursiveLlamaConfig.from_dict({**config.to_dict(), "recurse_layers": recurse_layers, "rank": rank, "alpha": alpha})
|
||||
config = RelaxedRecursiveLlamaConfig.from_dict(
|
||||
{
|
||||
**config.to_dict(),
|
||||
"recurse_layers": recurse_layers,
|
||||
"rank": rank,
|
||||
"alpha": alpha,
|
||||
"use_dora": use_dora,
|
||||
}
|
||||
)
|
||||
config.save_pretrained(output_dir)
|
||||
tokenizer.save_pretrained(output_dir)
|
||||
model_path = Path(snapshot_download(model_name, ignore_patterns="*.pth"))
|
||||
|
||||
# create a new state_dict to store the RRT model weights
|
||||
rrt_model_state_dict = {}
|
||||
|
||||
logger.info(f"Calculating average recursive weights...")
|
||||
for key, weight in iter_recursive_parameter_weights(model_path, modules_to_recurse, device=device, recurse_layers=recurse_layers):
|
||||
logger.info("Calculating average recursive weights...")
|
||||
for key, weight in iter_recursive_parameter_weights(
|
||||
model_path, modules_to_recurse, device=device, recurse_layers=recurse_layers
|
||||
):
|
||||
rrt_model_state_dict[key] = weight.to(torch.bfloat16).detach().cpu()
|
||||
|
||||
logger.info(f"Calculating decomposed lora diff...")
|
||||
logger.info("Calculating decomposed lora diff...")
|
||||
# now that we have the average weights, we need to loop over the shards again to calculate the decomposed lora diff
|
||||
rrt_lora_state_dict = {}
|
||||
for key, weight in iter_dora_parameter_weights(model_path, rrt_model_state_dict, modules_to_recurse, alpha=32, rank=rank, device=device, recurse_layers=recurse_layers):
|
||||
for key, weight in iter_dora_parameter_weights(
|
||||
model_path,
|
||||
rrt_model_state_dict,
|
||||
modules_to_recurse,
|
||||
alpha=32,
|
||||
rank=rank,
|
||||
device=device,
|
||||
recurse_layers=recurse_layers,
|
||||
use_dora=use_dora,
|
||||
):
|
||||
rrt_lora_state_dict[key] = weight.to(torch.bfloat16).detach().cpu()
|
||||
|
||||
# combine state dicts into a single state_dict
|
||||
@@ -279,10 +359,12 @@ def convert_llama_to_rrt(model_name, output_dir, recurse_layers: int = 12, rank=
|
||||
|
||||
if __name__ == "__main__":
|
||||
# meta-llama/Llama-3.2-1B has 16 hidden layers
|
||||
if torch.backends.mps.is_available():
|
||||
device = "mps"
|
||||
elif torch.cuda.is_available():
|
||||
device = "cuda"
|
||||
else:
|
||||
device = "cpu"
|
||||
convert_llama_to_rrt("meta-llama/Llama-3.2-1B", "/tmp/rrt_model", recurse_layers=4, rank=256, alpha=512, device=device)
|
||||
# meta-llama/Llama-3.2-3B has 28 hidden layers
|
||||
convert_llama_to_rrt(
|
||||
"meta-llama/Llama-3.2-3B",
|
||||
"/tmp/rrt_model", # nosec
|
||||
recurse_layers=4,
|
||||
rank=256,
|
||||
alpha=512,
|
||||
use_dora=False,
|
||||
)
|
||||
|
||||
@@ -1,2 +1,25 @@
|
||||
"""
|
||||
module for modeling relaxed recursive transformers model
|
||||
"""
|
||||
from transformers import AutoConfig, AutoModel, AutoModelForCausalLM
|
||||
|
||||
from .configuration_rrt_llama import RelaxedRecursiveLlamaConfig
|
||||
from .modeling_rrt_llama import (
|
||||
RelaxedRecursiveLlamaForCausalLM,
|
||||
RelaxedRecursiveLlamaModel,
|
||||
)
|
||||
|
||||
|
||||
def register_rrt_model():
|
||||
pass
|
||||
"""
|
||||
Register Relaxed Recursive Transformers model with transformers
|
||||
"""
|
||||
|
||||
# Register configs
|
||||
AutoConfig.register("llama-rrt", RelaxedRecursiveLlamaConfig)
|
||||
|
||||
# Register models
|
||||
AutoModel.register(RelaxedRecursiveLlamaConfig, RelaxedRecursiveLlamaModel)
|
||||
AutoModelForCausalLM.register(
|
||||
RelaxedRecursiveLlamaConfig, RelaxedRecursiveLlamaForCausalLM
|
||||
)
|
||||
|
||||
@@ -0,0 +1,16 @@
|
||||
"""
|
||||
module for custom configuration for relaxed recursive transformers model
|
||||
"""
|
||||
from transformers import LlamaConfig
|
||||
|
||||
|
||||
class RelaxedRecursiveLlamaConfig(LlamaConfig):
|
||||
"""
|
||||
Configuration for Relaxed Recursive Llama.
|
||||
"""
|
||||
|
||||
model_type: str = "llama-rrt"
|
||||
recurse_layers: int = 4
|
||||
rank: int
|
||||
alpha: int
|
||||
use_dora: bool = True
|
||||
@@ -1,3 +1,6 @@
|
||||
"""
|
||||
module for the shared linear layer for the relaxed recursive transformers model
|
||||
"""
|
||||
import math
|
||||
|
||||
import torch
|
||||
@@ -6,7 +9,6 @@ from peft.utils import transpose
|
||||
from torch import nn
|
||||
|
||||
|
||||
|
||||
class RelaxedRecursiveDoraLinear(nn.Module):
|
||||
"""
|
||||
A single linear layer that is "shared" across multiple loop iterations,
|
||||
@@ -25,7 +27,7 @@ class RelaxedRecursiveDoraLinear(nn.Module):
|
||||
self,
|
||||
in_features: int,
|
||||
out_features: int,
|
||||
B: int,
|
||||
B: int, # pylint: disable=invalid-name
|
||||
rank: int,
|
||||
alpha: int,
|
||||
fan_in_fan_out: bool = False,
|
||||
@@ -33,7 +35,7 @@ class RelaxedRecursiveDoraLinear(nn.Module):
|
||||
use_dora: bool = True,
|
||||
):
|
||||
super().__init__()
|
||||
self.B = B
|
||||
self.B = B # pylint: disable=invalid-name
|
||||
self.fan_in_fan_out = fan_in_fan_out
|
||||
|
||||
self.weight_base = nn.Parameter(torch.empty(out_features, in_features))
|
||||
@@ -44,13 +46,19 @@ class RelaxedRecursiveDoraLinear(nn.Module):
|
||||
else:
|
||||
self.register_parameter("bias", None)
|
||||
|
||||
self.lora_A_list = nn.ParameterList([nn.Parameter(torch.zeros(rank, in_features)) for _ in range(B)])
|
||||
self.lora_B_list = nn.ParameterList([nn.Parameter(torch.zeros(out_features, rank)) for _ in range(B)])
|
||||
self.lora_A_list = nn.ParameterList( # pylint: disable=invalid-name
|
||||
[nn.Parameter(torch.zeros(rank, in_features)) for _ in range(B)]
|
||||
)
|
||||
self.lora_B_list = nn.ParameterList( # pylint: disable=invalid-name
|
||||
[nn.Parameter(torch.zeros(out_features, rank)) for _ in range(B)]
|
||||
)
|
||||
# rslora
|
||||
self.scaling = alpha / math.sqrt(rank)
|
||||
self.use_dora = use_dora
|
||||
if use_dora:
|
||||
self.lora_magnitude_vector_list = nn.ParameterList([nn.Parameter(torch.ones(out_features)) for _ in range(B)])
|
||||
self.lora_magnitude_vector_list = nn.ParameterList(
|
||||
[nn.Parameter(torch.ones(out_features)) for _ in range(B)]
|
||||
)
|
||||
|
||||
def get_weight_norm(self, weight, lora_weight, scaling) -> torch.Tensor:
|
||||
# calculate L2 norm of weight matrix, column-wise
|
||||
@@ -66,27 +74,43 @@ class RelaxedRecursiveDoraLinear(nn.Module):
|
||||
:param loop_idx:
|
||||
:return:
|
||||
"""
|
||||
eps = 1e-6
|
||||
w_base = self.weight_base
|
||||
w_base = w_base.to(x.dtype)
|
||||
|
||||
lora_A: torch.Tensor = self.lora_A_list[loop_idx]
|
||||
lora_B: torch.Tensor = self.lora_B_list[loop_idx]
|
||||
lora_A: torch.Tensor = self.lora_A_list[ # pylint: disable=invalid-name
|
||||
loop_idx
|
||||
]
|
||||
lora_B: torch.Tensor = self.lora_B_list[ # pylint: disable=invalid-name
|
||||
loop_idx
|
||||
]
|
||||
|
||||
base_out: torch.Tensor = F.linear(x, w_base, self.bias)
|
||||
|
||||
x_eye: torch.Tensor = torch.eye(lora_A.shape[1], device=lora_A.device, dtype=x.dtype)
|
||||
tmp = F.linear(x_eye, lora_A) # [hidden_size, rank]
|
||||
w_dora_full: torch.Tensor = F.linear(tmp, lora_B)
|
||||
w_dora_full = w_dora_full.t()
|
||||
|
||||
lora_out: torch.Tensor = F.linear(x, w_dora_full, bias=None)
|
||||
lora_out: torch.Tensor = F.linear(F.linear(x, lora_A), lora_B) * self.scaling
|
||||
|
||||
if self.use_dora:
|
||||
x_eye: torch.Tensor = torch.eye(
|
||||
lora_A.shape[1], device=lora_A.device, dtype=x.dtype
|
||||
)
|
||||
tmp = F.linear(x_eye, lora_A) # [hidden_size, rank]
|
||||
w_dora_full: torch.Tensor = F.linear(tmp, lora_B)
|
||||
w_dora_full = w_dora_full.t()
|
||||
|
||||
magnitude_vector: torch.Tensor = self.lora_magnitude_vector_list[loop_idx]
|
||||
w_dora_norm: torch.Tensor = self.get_weight_norm(w_base, w_dora_full.detach(), self.scaling)
|
||||
w_dora_norm: torch.Tensor = self.get_weight_norm(
|
||||
w_base, w_dora_full.detach(), self.scaling
|
||||
)
|
||||
w_dora_norm = w_dora_norm.detach()
|
||||
scale_factor = (magnitude_vector / w_dora_norm).unsqueeze(0) # shape [1, out_features]
|
||||
scale_factor = (magnitude_vector / w_dora_norm).unsqueeze(
|
||||
0
|
||||
) # shape [1, out_features]
|
||||
|
||||
result_dora = (scale_factor - 1) * base_out + scale_factor * lora_out
|
||||
return result_dora
|
||||
return base_out + lora_out * self.scaling
|
||||
|
||||
# scale the lora norm to prevent gradient explosion
|
||||
orig_norm = torch.linalg.norm(w_base)
|
||||
update_norm = torch.linalg.norm(lora_out)
|
||||
scale = orig_norm / (update_norm + eps)
|
||||
|
||||
return base_out + lora_out * scale
|
||||
|
||||
@@ -1,30 +1,31 @@
|
||||
import logging
|
||||
from typing import Tuple, Optional, Unpack, Callable, Union
|
||||
from typing import Callable, Optional, Tuple, Union, Unpack
|
||||
|
||||
import torch
|
||||
from torch import nn
|
||||
from transformers import LlamaConfig, Cache, DynamicCache
|
||||
from transformers import Cache, DynamicCache, LlamaConfig
|
||||
from transformers.activations import ACT2FN
|
||||
from transformers.modeling_flash_attention_utils import FlashAttentionKwargs
|
||||
from transformers.modeling_outputs import BaseModelOutputWithPast
|
||||
from transformers.modeling_utils import ALL_ATTENTION_FUNCTIONS
|
||||
from transformers.models.llama.modeling_llama import apply_rotary_pos_emb, eager_attention_forward, LlamaRMSNorm, \
|
||||
LlamaForCausalLM, LlamaModel, LlamaRotaryEmbedding
|
||||
from transformers.models.llama.modeling_llama import (
|
||||
LlamaForCausalLM,
|
||||
LlamaModel,
|
||||
LlamaRMSNorm,
|
||||
LlamaRotaryEmbedding,
|
||||
apply_rotary_pos_emb,
|
||||
eager_attention_forward,
|
||||
)
|
||||
|
||||
from axolotl.integrations.rrt.modeling.linear import RelaxedRecursiveDoraLinear
|
||||
|
||||
from .configuration_rrt_llama import RelaxedRecursiveLlamaConfig
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
class RelaxedRecursiveLlamaConfig(LlamaConfig):
|
||||
"""
|
||||
Configuration for Relaxed Recursive Llama.
|
||||
"""
|
||||
|
||||
model_type = "llama-rrt"
|
||||
recurse_layers: int = 4
|
||||
rank: int
|
||||
alpha: int
|
||||
use_dora: bool = True
|
||||
# pylint: skip-file
|
||||
# mypy: ignore-errors
|
||||
|
||||
|
||||
class RelaxedRecursiveLlamaMLP(nn.Module):
|
||||
@@ -34,13 +35,40 @@ class RelaxedRecursiveLlamaMLP(nn.Module):
|
||||
self.config = config
|
||||
self.hidden_size = config.hidden_size
|
||||
self.intermediate_size = config.intermediate_size
|
||||
self.gate_proj = RelaxedRecursiveDoraLinear(self.hidden_size, self.intermediate_size, recurse_loops, config.rank, config.alpha, bias=config.mlp_bias, use_dora=config.use_dora)
|
||||
self.up_proj = RelaxedRecursiveDoraLinear(self.hidden_size, self.intermediate_size, recurse_loops, config.rank, config.alpha, bias=config.mlp_bias, use_dora=config.use_dora)
|
||||
self.down_proj = RelaxedRecursiveDoraLinear(self.intermediate_size, self.hidden_size, recurse_loops, config.rank, config.alpha, bias=config.mlp_bias, use_dora=config.use_dora)
|
||||
self.gate_proj = RelaxedRecursiveDoraLinear(
|
||||
self.hidden_size,
|
||||
self.intermediate_size,
|
||||
recurse_loops,
|
||||
config.rank,
|
||||
config.alpha,
|
||||
bias=config.mlp_bias,
|
||||
use_dora=config.use_dora,
|
||||
)
|
||||
self.up_proj = RelaxedRecursiveDoraLinear(
|
||||
self.hidden_size,
|
||||
self.intermediate_size,
|
||||
recurse_loops,
|
||||
config.rank,
|
||||
config.alpha,
|
||||
bias=config.mlp_bias,
|
||||
use_dora=config.use_dora,
|
||||
)
|
||||
self.down_proj = RelaxedRecursiveDoraLinear(
|
||||
self.intermediate_size,
|
||||
self.hidden_size,
|
||||
recurse_loops,
|
||||
config.rank,
|
||||
config.alpha,
|
||||
bias=config.mlp_bias,
|
||||
use_dora=config.use_dora,
|
||||
)
|
||||
self.act_fn = ACT2FN[config.hidden_act]
|
||||
|
||||
def forward(self, x, loop_idx: int):
|
||||
down_proj = self.down_proj(self.act_fn(self.gate_proj(x, loop_idx)) * self.up_proj(x, loop_idx), loop_idx)
|
||||
down_proj = self.down_proj(
|
||||
self.act_fn(self.gate_proj(x, loop_idx)) * self.up_proj(x, loop_idx),
|
||||
loop_idx,
|
||||
)
|
||||
return down_proj
|
||||
|
||||
|
||||
@@ -54,23 +82,51 @@ class RelaxedRecursiveLlamaAttention(nn.Module):
|
||||
recurse_loops = config.num_hidden_layers // config.recurse_layers
|
||||
self.config = config
|
||||
self.layer_idx = layer_idx
|
||||
self.head_dim = getattr(config, "head_dim", config.hidden_size // config.num_attention_heads)
|
||||
self.num_key_value_groups = config.num_attention_heads // config.num_key_value_heads
|
||||
self.head_dim = getattr(
|
||||
config, "head_dim", config.hidden_size // config.num_attention_heads
|
||||
)
|
||||
self.num_key_value_groups = (
|
||||
config.num_attention_heads // config.num_key_value_heads
|
||||
)
|
||||
self.scaling = self.head_dim**-0.5
|
||||
self.attention_dropout = config.attention_dropout
|
||||
self.is_causal = True
|
||||
|
||||
self.q_proj = RelaxedRecursiveDoraLinear(
|
||||
config.hidden_size, config.num_attention_heads * self.head_dim, recurse_loops, config.rank, config.alpha, bias=config.attention_bias, use_dora=config.use_dora
|
||||
config.hidden_size,
|
||||
config.num_attention_heads * self.head_dim,
|
||||
recurse_loops,
|
||||
config.rank,
|
||||
config.alpha,
|
||||
bias=config.attention_bias,
|
||||
use_dora=config.use_dora,
|
||||
)
|
||||
self.k_proj = RelaxedRecursiveDoraLinear(
|
||||
config.hidden_size, config.num_key_value_heads * self.head_dim, recurse_loops, config.rank, config.alpha, bias=config.attention_bias, use_dora=config.use_dora
|
||||
config.hidden_size,
|
||||
config.num_key_value_heads * self.head_dim,
|
||||
recurse_loops,
|
||||
config.rank,
|
||||
config.alpha,
|
||||
bias=config.attention_bias,
|
||||
use_dora=config.use_dora,
|
||||
)
|
||||
self.v_proj = RelaxedRecursiveDoraLinear(
|
||||
config.hidden_size, config.num_key_value_heads * self.head_dim, recurse_loops, config.rank, config.alpha, bias=config.attention_bias, use_dora=config.use_dora
|
||||
config.hidden_size,
|
||||
config.num_key_value_heads * self.head_dim,
|
||||
recurse_loops,
|
||||
config.rank,
|
||||
config.alpha,
|
||||
bias=config.attention_bias,
|
||||
use_dora=config.use_dora,
|
||||
)
|
||||
self.o_proj = RelaxedRecursiveDoraLinear(
|
||||
config.num_attention_heads * self.head_dim, config.hidden_size, recurse_loops, config.rank, config.alpha, bias=config.attention_bias, use_dora=config.use_dora
|
||||
config.num_attention_heads * self.head_dim,
|
||||
config.hidden_size,
|
||||
recurse_loops,
|
||||
config.rank,
|
||||
config.alpha,
|
||||
bias=config.attention_bias,
|
||||
use_dora=config.use_dora,
|
||||
)
|
||||
|
||||
def forward(
|
||||
@@ -81,32 +137,46 @@ class RelaxedRecursiveLlamaAttention(nn.Module):
|
||||
loop_idx: int,
|
||||
past_key_value: Optional[Cache] = None,
|
||||
cache_position: Optional[torch.LongTensor] = None,
|
||||
**kwargs: Unpack[FlashAttentionKwargs],
|
||||
**kwargs: Unpack[FlashAttentionKwargs], # pylint: disable=misc
|
||||
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
|
||||
input_shape = hidden_states.shape[:-1]
|
||||
hidden_shape = (*input_shape, -1, self.head_dim)
|
||||
|
||||
query_states = self.q_proj(hidden_states, loop_idx).view(hidden_shape).transpose(1, 2)
|
||||
key_states = self.k_proj(hidden_states, loop_idx).view(hidden_shape).transpose(1, 2)
|
||||
value_states = self.v_proj(hidden_states, loop_idx).view(hidden_shape).transpose(1, 2)
|
||||
query_states = (
|
||||
self.q_proj(hidden_states, loop_idx).view(hidden_shape).transpose(1, 2)
|
||||
)
|
||||
key_states = (
|
||||
self.k_proj(hidden_states, loop_idx).view(hidden_shape).transpose(1, 2)
|
||||
)
|
||||
value_states = (
|
||||
self.v_proj(hidden_states, loop_idx).view(hidden_shape).transpose(1, 2)
|
||||
)
|
||||
|
||||
cos, sin = position_embeddings
|
||||
query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin)
|
||||
query_states, key_states = apply_rotary_pos_emb(
|
||||
query_states, key_states, cos, sin
|
||||
)
|
||||
|
||||
if past_key_value is not None:
|
||||
# sin and cos are specific to RoPE models; cache_position needed for the static cache
|
||||
cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position}
|
||||
key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
|
||||
key_states, value_states = past_key_value.update(
|
||||
key_states, value_states, self.layer_idx, cache_kwargs
|
||||
)
|
||||
|
||||
attention_interface: Callable = eager_attention_forward
|
||||
if self.config._attn_implementation != "eager":
|
||||
if self.config._attn_implementation == "sdpa" and kwargs.get("output_attentions", False):
|
||||
logger.warning_once(
|
||||
if self.config._attn_implementation == "sdpa" and kwargs.get(
|
||||
"output_attentions", False
|
||||
):
|
||||
logger.warning(
|
||||
"`torch.nn.functional.scaled_dot_product_attention` does not support `output_attentions=True`. Falling back to "
|
||||
'eager attention. This warning can be removed using the argument `attn_implementation="eager"` when loading the model.'
|
||||
)
|
||||
else:
|
||||
attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation]
|
||||
attention_interface = ALL_ATTENTION_FUNCTIONS[
|
||||
self.config._attn_implementation
|
||||
]
|
||||
|
||||
attn_output, attn_weights = attention_interface(
|
||||
self,
|
||||
@@ -121,8 +191,7 @@ class RelaxedRecursiveLlamaAttention(nn.Module):
|
||||
|
||||
attn_output = attn_output.reshape(*input_shape, -1).contiguous()
|
||||
attn_output = self.o_proj(attn_output, loop_idx)
|
||||
return attn_output, attn_weights
|
||||
|
||||
return attn_output, attn_weights # pylint: disable=return-value
|
||||
|
||||
|
||||
class RelaxedRecursiveLlamaDecoderLayer(nn.Module):
|
||||
@@ -135,12 +204,24 @@ class RelaxedRecursiveLlamaDecoderLayer(nn.Module):
|
||||
recurse_loops = config.num_hidden_layers // config.recurse_layers
|
||||
self.hidden_size = config.hidden_size
|
||||
|
||||
self.self_attn = RelaxedRecursiveLlamaAttention(config=config, layer_idx=layer_idx)
|
||||
self.self_attn = RelaxedRecursiveLlamaAttention(
|
||||
config=config, layer_idx=layer_idx
|
||||
)
|
||||
|
||||
self.mlp = RelaxedRecursiveLlamaMLP(config)
|
||||
|
||||
self.input_layernorm_list = nn.ModuleList([LlamaRMSNorm(config.hidden_size, eps=config.rms_norm_eps) for _ in range(recurse_loops)])
|
||||
self.post_attention_layernorm_list = nn.ModuleList([LlamaRMSNorm(config.hidden_size, eps=config.rms_norm_eps) for _ in range(recurse_loops)])
|
||||
self.input_layernorm_list = nn.ModuleList(
|
||||
[
|
||||
LlamaRMSNorm(config.hidden_size, eps=config.rms_norm_eps)
|
||||
for _ in range(recurse_loops)
|
||||
]
|
||||
)
|
||||
self.post_attention_layernorm_list = nn.ModuleList(
|
||||
[
|
||||
LlamaRMSNorm(config.hidden_size, eps=config.rms_norm_eps)
|
||||
for _ in range(recurse_loops)
|
||||
]
|
||||
)
|
||||
|
||||
def forward(
|
||||
self,
|
||||
@@ -152,9 +233,13 @@ class RelaxedRecursiveLlamaDecoderLayer(nn.Module):
|
||||
output_attentions: Optional[bool] = False,
|
||||
use_cache: Optional[bool] = False,
|
||||
cache_position: Optional[torch.LongTensor] = None,
|
||||
position_embeddings: Optional[Tuple[torch.Tensor, torch.Tensor]] = None, # necessary, but kept here for BC
|
||||
**kwargs: Unpack[FlashAttentionKwargs],
|
||||
) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
|
||||
position_embeddings: Optional[
|
||||
Tuple[torch.Tensor, torch.Tensor]
|
||||
] = None, # necessary, but kept here for BC
|
||||
**kwargs: Unpack[FlashAttentionKwargs], # pylint: disable=misc
|
||||
) -> Tuple[
|
||||
torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]
|
||||
]:
|
||||
residual = hidden_states
|
||||
|
||||
hidden_states = self.input_layernorm_list[loop_idx](hidden_states)
|
||||
@@ -196,9 +281,14 @@ class RelaxedRecursiveLlamaModel(LlamaModel):
|
||||
self.padding_idx = config.pad_token_id
|
||||
self.vocab_size = config.vocab_size
|
||||
|
||||
self.embed_tokens = nn.Embedding(config.vocab_size, config.hidden_size, self.padding_idx)
|
||||
self.embed_tokens = nn.Embedding(
|
||||
config.vocab_size, config.hidden_size, self.padding_idx
|
||||
)
|
||||
self.layers = nn.ModuleList(
|
||||
[RelaxedRecursiveLlamaDecoderLayer(config, layer_idx) for layer_idx in range(config.recurse_layers)]
|
||||
[
|
||||
RelaxedRecursiveLlamaDecoderLayer(config, layer_idx)
|
||||
for layer_idx in range(config.recurse_layers)
|
||||
]
|
||||
)
|
||||
self.norm = LlamaRMSNorm(config.hidden_size, eps=config.rms_norm_eps)
|
||||
self.rotary_emb = LlamaRotaryEmbedding(config=config)
|
||||
@@ -221,15 +311,25 @@ class RelaxedRecursiveLlamaModel(LlamaModel):
|
||||
cache_position: Optional[torch.LongTensor] = None,
|
||||
**flash_attn_kwargs: Unpack[FlashAttentionKwargs],
|
||||
) -> Union[Tuple, BaseModelOutputWithPast]:
|
||||
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
|
||||
output_attentions = (
|
||||
output_attentions
|
||||
if output_attentions is not None
|
||||
else self.config.output_attentions
|
||||
)
|
||||
output_hidden_states = (
|
||||
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
|
||||
output_hidden_states
|
||||
if output_hidden_states is not None
|
||||
else self.config.output_hidden_states
|
||||
)
|
||||
use_cache = use_cache if use_cache is not None else self.config.use_cache
|
||||
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
|
||||
return_dict = (
|
||||
return_dict if return_dict is not None else self.config.use_return_dict
|
||||
)
|
||||
|
||||
if (input_ids is None) ^ (inputs_embeds is not None):
|
||||
raise ValueError("You must specify exactly one of input_ids or inputs_embeds")
|
||||
raise ValueError(
|
||||
"You must specify exactly one of input_ids or inputs_embeds"
|
||||
)
|
||||
|
||||
if self.gradient_checkpointing and self.training and use_cache:
|
||||
logger.warning_once(
|
||||
@@ -244,16 +344,24 @@ class RelaxedRecursiveLlamaModel(LlamaModel):
|
||||
past_key_values = DynamicCache()
|
||||
|
||||
if cache_position is None:
|
||||
past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0
|
||||
past_seen_tokens = (
|
||||
past_key_values.get_seq_length() if past_key_values is not None else 0
|
||||
)
|
||||
cache_position = torch.arange(
|
||||
past_seen_tokens, past_seen_tokens + inputs_embeds.shape[1], device=inputs_embeds.device
|
||||
past_seen_tokens,
|
||||
past_seen_tokens + inputs_embeds.shape[1],
|
||||
device=inputs_embeds.device,
|
||||
)
|
||||
|
||||
if position_ids is None:
|
||||
position_ids = cache_position.unsqueeze(0)
|
||||
|
||||
causal_mask = self._update_causal_mask(
|
||||
attention_mask, inputs_embeds, cache_position, past_key_values, output_attentions
|
||||
attention_mask,
|
||||
inputs_embeds,
|
||||
cache_position,
|
||||
past_key_values,
|
||||
output_attentions,
|
||||
)
|
||||
|
||||
hidden_states = inputs_embeds
|
||||
|
||||
Reference in New Issue
Block a user