Coverage for transformer_lens/components/mlps/gated_mlp_4bit.py: 42%

1"""Hooked Transformer Gated MLP Component. 

2 

3This module contains all the component :class:`GatedMLP`. 

4""" 

5 

6from typing import Dict, Union 

7 

8import torch 

9import torch.nn as nn 

10from jaxtyping import Float 

11from transformers.utils import is_bitsandbytes_available 

12 

13from transformer_lens.components.mlps.can_be_used_as_mlp import CanBeUsedAsMLP 

14from transformer_lens.hook_points import HookPoint 

15from transformer_lens.HookedTransformerConfig import HookedTransformerConfig 

16 

17if is_bitsandbytes_available(): 17 ↛ 18line 17 didn't jump to line 18 because the condition on line 17 was never true

18 import bitsandbytes as bnb 

19 from bitsandbytes.nn.modules import Params4bit 

20 

21 

22class GatedMLP4Bit(CanBeUsedAsMLP): 

23 """ 

24 The equation of a gated MLP: 

25 pre = x @ W_gate 

26 pre_linear = x @ W_in 

27 post = Gelu(pre) * (pre_linear) + b_in 

28 mlp_out = post @ W_out + b_out 

29 

30 In one equation, mlp_out = (Gelu(x @ W_gate) * (x @ W_in) + b_in) @ W_out + b_out 

31 """ 

32 

33 def __init__(self, cfg: Union[Dict, HookedTransformerConfig]): 

34 super().__init__(cfg) 

35 self.select_activation_function() 

36 

37 nq = int((self.cfg.d_model * self.d_mlp) / 2) 

38 self.W_in = Params4bit(torch.empty(nq, 1, dtype=torch.uint8), requires_grad=False) 

39 self.W_gate = Params4bit(torch.empty(nq, 1, dtype=torch.uint8), requires_grad=False) 

40 self.W_out = Params4bit(torch.empty(nq, 1, dtype=torch.uint8), requires_grad=False) 

41 

42 self.b_in = nn.Parameter(torch.zeros(self.d_mlp, dtype=self.cfg.dtype)) 

43 self.b_out = nn.Parameter(torch.zeros(self.cfg.d_model, dtype=self.cfg.dtype)) 

44 

45 # hook on gate output but before act_fn 

46 self.hook_pre = HookPoint() # [batch, pos, d_mlp] 

47 # hook on the linear component of the input 

48 self.hook_pre_linear = HookPoint() # [batch, pos, d_mlp] 

49 # hook on act_fn(gate_output) * W_in(x) + b_in 

50 self.hook_post = HookPoint() # [batch, pos, d_mlp] 

51 

52 def forward( 

53 self, x: Float[torch.Tensor, "batch pos d_model"] 

54 ) -> Float[torch.Tensor, "batch pos d_model"]: 

55 # Technically, all these einsums could be done with a single matmul, but this is more readable. 

56 pre_act = self.hook_pre( 

57 bnb.matmul_4bit(x, self.W_gate.t(), bias=None, quant_state=self.W_gate.quant_state) 

58 ) 

59 

60 if ( 

61 self.cfg.is_layer_norm_activation() 

62 and self.hook_mid is not None 

63 and self.ln is not None 

64 ): 

65 mid_act = self.hook_mid(self.act_fn(pre_act)) # [batch, pos, d_mlp] 

66 post_act = self.hook_post(self.ln(mid_act)) 

67 else: 

68 pre_linear = self.hook_pre_linear( 

69 bnb.matmul_4bit(x, self.W_in.t(), bias=None, quant_state=self.W_in.quant_state) 

70 ) 

71 

72 post_act = self.hook_post( 

73 (self.act_fn(pre_act) * pre_linear) + self.b_in 

74 ) # [batch, pos, d_mlp] 

75 

76 return bnb.matmul_4bit( 

77 post_act, self.W_out.t(), bias=None, quant_state=self.W_out.quant_state 

78 )