Upload folder using huggingface_hub
Browse files
README.md
CHANGED
@@ -6,7 +6,7 @@ tags:
|
|
6 |
|
7 |
# LagKV Cache
|
8 |
|
9 |
-
|
10 |
|
11 |

|
12 |
|
@@ -18,99 +18,34 @@ Details are in the following work:
|
|
18 |
|
19 |
[LagKV: Lag-Relative Information of the KV Cache Tells Which Tokens Are Important](https://arxiv.org/abs/2504.04704)
|
20 |
|
21 |
-
|
22 |
|
23 |
-
|
24 |
|
25 |
-
```
|
26 |
-
|
27 |
from transformers import AutoModelForCausalLM, AutoTokenizer
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
lag_size=128,
|
53 |
-
layer_idx_skip_first=[],
|
54 |
-
use_then_compress=True)
|
55 |
-
|
56 |
-
with torch.no_grad():
|
57 |
-
sink_size = past_key_values.sink_size
|
58 |
-
lag_size = past_key_values.lag_size
|
59 |
-
trigger_len = sink_size + 2*lag_size
|
60 |
-
input_length = input_ids.shape[1]
|
61 |
-
# print(input_length > trigger_len)
|
62 |
-
if input_length > trigger_len:
|
63 |
-
start_idx = 0
|
64 |
-
end_idx = trigger_len
|
65 |
-
position_ids = torch.arange(input_length + max_new_tokens).unsqueeze(0).to(device)
|
66 |
-
def batch_input():
|
67 |
-
sel_input_ids = input_ids[:, start_idx:end_idx]
|
68 |
-
q_len = end_idx - start_idx
|
69 |
-
k_len = past_key_values.get_seq_length() + q_len
|
70 |
-
batch_size = input_ids.shape[0]
|
71 |
-
head_num = model.config.num_attention_heads
|
72 |
-
attn_mask = torch.ones((k_len, q_len),
|
73 |
-
device=input_ids.device, dtype=torch.bool)
|
74 |
-
attn_mask = torch.triu(attn_mask, diagonal=1).T
|
75 |
-
attn_mask = torch.flip(attn_mask, (0, 1))
|
76 |
-
attn_mask = attn_mask.unsqueeze(0).unsqueeze(0)
|
77 |
-
attn_mask = attn_mask.expand(batch_size, -1, -1, -1).expand(-1, head_num, -1, -1)
|
78 |
-
attention_mask = torch.zeros((batch_size, head_num, q_len, k_len), device=input_ids.device, dtype=torch.bfloat16)
|
79 |
-
attention_mask.masked_fill_(attn_mask, -torch.inf)
|
80 |
-
return {"input_ids": sel_input_ids, "attention_mask": attention_mask}
|
81 |
-
|
82 |
-
while start_idx < input_length:
|
83 |
-
tmp_pos = position_ids[:, start_idx:end_idx]
|
84 |
-
outputs = model(**batch_input(),
|
85 |
-
past_key_values=past_key_values,
|
86 |
-
position_ids=tmp_pos,
|
87 |
-
cache_position=tmp_pos[0]
|
88 |
-
)
|
89 |
-
start_idx = end_idx
|
90 |
-
end_idx += lag_size
|
91 |
-
end_idx = min(end_idx, input_length)
|
92 |
-
|
93 |
-
new_token_id = outputs.logits[:, -1].argmax(dim=-1).unsqueeze(-1)
|
94 |
-
# print(new_token_id)
|
95 |
-
new_token_count = 1
|
96 |
-
generated_ids = [new_token_id]
|
97 |
-
while new_token_id[0][0] != tokenizer.eos_token_id and new_token_count < max_new_tokens+1:
|
98 |
-
tmp_pos = position_ids[:, (input_length+new_token_count-1):(input_length+new_token_count)]
|
99 |
-
outputs = model(new_token_id,
|
100 |
-
past_key_values=past_key_values,
|
101 |
-
position_ids=tmp_pos,
|
102 |
-
cache_position=tmp_pos[0]
|
103 |
-
)
|
104 |
-
new_token_id = outputs.logits[:, -1].argmax(dim=-1).unsqueeze(-1)
|
105 |
-
new_token_count += 1
|
106 |
-
generated_ids.append(new_token_id)
|
107 |
-
generated_ids = torch.cat(generated_ids, dim=-1)
|
108 |
-
else:
|
109 |
-
generated_ids = model.generate(inputs, do_sample=False, max_new_tokens=max_new_tokens, past_key_values=past_key_values)
|
110 |
-
generated_ids = generated_ids[:, input_length:]
|
111 |
-
if decode:
|
112 |
-
output = tokenizer.batch_decode(generated_ids)
|
113 |
-
else:
|
114 |
-
output = generated_ids
|
115 |
-
return output, past_key_values
|
116 |
-
```
|
|
|
6 |
|
7 |
# LagKV Cache
|
8 |
|
9 |
+
## Introduction
|
10 |
|
11 |

|
12 |
|
|
|
18 |
|
19 |
[LagKV: Lag-Relative Information of the KV Cache Tells Which Tokens Are Important](https://arxiv.org/abs/2504.04704)
|
20 |
|
21 |
+
## Example usage
|
22 |
|
23 |
+
We can use the custom generation method in this repository like the the base `generate` from `transformers`:
|
24 |
|
25 |
+
```py
|
26 |
+
# requires `transformers>=4.52.0`
|
27 |
from transformers import AutoModelForCausalLM, AutoTokenizer
|
28 |
+
# Preparing model, tokenizer, and model inputs
|
29 |
+
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen3-0.6B")
|
30 |
+
model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-0.6B", device_map="auto")
|
31 |
+
messages = [{"role": "user", "content": "Tell me a story about a cat."}]
|
32 |
+
text = tokenizer.apply_chat_template(
|
33 |
+
messages,
|
34 |
+
tokenize=False,
|
35 |
+
add_generation_prompt=True,
|
36 |
+
enable_thinking=False
|
37 |
+
)
|
38 |
+
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
|
39 |
+
# Using lagkv cache
|
40 |
+
gen_out = model.generate(
|
41 |
+
# usual `generate` arguments
|
42 |
+
**model_inputs,
|
43 |
+
do_sample=False,
|
44 |
+
max_new_tokens=100,
|
45 |
+
return_dict_in_generate=True,
|
46 |
+
# lagkv cache arguments (default `lag_ratio=0.5,lag_size=128,lag_sink_size=16`)
|
47 |
+
custom_generate="CMB-AI-LAB/lagkv_cache",
|
48 |
+
trust_remote_code=True,
|
49 |
+
)
|
50 |
+
print(tokenizer.batch_decode(gen_out.sequences, skip_special_tokens=True))
|
51 |
+
assert "lagkvcache" in str(type(gen_out.past_key_values)).lower()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|