ceos/unsloth
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Update llama.py

Browse source
This commit is contained in:
Daniel Han-Chen 2024-02-01 18:16:15 +11:00
commit 3f0ddf0c7c
1 changed files with 15 additions and 15 deletions
Download patch file Download diff file Expand all files Collapse all files

26
unsloth/models/llama.py
View file

@ -230,7 +230,7 @@ def LlamaAttention_fast_forward_inference(
n_groups = self.num_key_value_groups
n_kv_heads = self.num_key_value_heads
head_dim = self.head_dim
assert(n_kv_heads * n_groups == n_heads)
# ssert(n_kv_heads * n_groups == n_heads)
Qn = fast_linear_forward(self.q_proj, Xn)
Kn = fast_linear_forward(self.k_proj, Xn)
@ -241,19 +241,19 @@ def LlamaAttention_fast_forward_inference(
seq_len = K1.shape[-2]
kv_seq_len = seq_len + 1
# cos, sin = self.rotary_emb(Vn, seq_len = kv_seq_len)
# Qn, Kn = inplace_rope_embedding(Qn, Kn, cos, sin, position_ids)
cos = self.rotary_emb.cos_cached[seq_len]
sin = self.rotary_emb.sin_cached[seq_len]
h = head_dim // 2
cos, sin = self.rotary_emb(Vn, seq_len = kv_seq_len)
Qn, Kn = inplace_rope_embedding(Qn, Kn, cos, sin, position_ids)
# cos = self.rotary_emb.cos_cached[seq_len]
# sin = self.rotary_emb.sin_cached[seq_len]
# h = head_dim // 2
RH_Q = torch.empty((bsz, n_heads, 1, head_dim), dtype = Xn.dtype, device = "cuda")
RH_Q[:,:,:,:h] = Qn[:,:,:,h:]; RH_Q[:,:,:,h:] = Qn[:,:,:,:h]; torch.neg(RH_Q[:,:,:,:h], out = RH_Q[:,:,:,:h]);
Qn *= cos; Qn.addcmul_(RH_Q, sin);
# RH_Q = torch.empty((bsz, n_heads, 1, head_dim), dtype = Xn.dtype, device = "cuda")
# RH_Q[:,:,:,:h] = Qn[:,:,:,h:]; RH_Q[:,:,:,h:] = Qn[:,:,:,:h]; torch.neg(RH_Q[:,:,:,:h], out = RH_Q[:,:,:,:h]);
# Qn *= cos; Qn.addcmul_(RH_Q, sin);
RH_K = RH_Q[:,:n_kv_heads,:,:] # torch.empty((n_kv_heads, 1, head_dim), dtype = dtype, device = "cuda")
RH_K[:,:,:,:h] = Kn[:,:,:,h:]; RH_K[:,:,:,h:] = Kn[:,:,:,:h]; torch.neg(RH_K[:,:,:,:h], out = RH_K[:,:,:,:h]);
Kn *= cos; Kn.addcmul_(RH_K, sin);
# RH_K = RH_Q[:,:n_kv_heads,:,:] # torch.empty((n_kv_heads, 1, head_dim), dtype = dtype, device = "cuda")
# RH_K[:,:,:,:h] = Kn[:,:,:,h:]; RH_K[:,:,:,h:] = Kn[:,:,:,:h]; torch.neg(RH_K[:,:,:,:h], out = RH_K[:,:,:,:h]);
# Kn *= cos; Kn.addcmul_(RH_K, sin);
# New KV cache
Kn = torch.cat([K1, Kn], dim = 2)
@ -290,7 +290,7 @@ def fast_mlp_inference(self, X):
gate *= up
# X = self.down_proj(gate)
down = fast_linear_forward(self.down_proj, gate)
down = fast_linear_forward(self.down_proj, gate, out = up[:,:,:X.shape[2]])
return down
pass