File size: 2,521 Bytes
e647259
 
5a0aa6b
 
 
 
e647259
5a0aa6b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8d0b89e
5a0aa6b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
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
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
---
license: mit
language:
- en
base_model:
- LiquidAI/LFM2.5-230M
---
## LFM 2.5 230m Code-Enhanced
Lfm 2.5 230m code enhanced is a code-ish version of the previous lfm 2.5 230m (base) , this model however attempts to make the old model better at coding
this uses the latest weight space programming paper along with deepseek v4.1 flash to manually change the model weights.

<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>LFM CodeEnhanced AI</title>
<style>
* { box-sizing: border-box; }
body {
  margin: 0;
  min-height: 100vh;
  display: grid;
  place-items: center;
  background: #080b14;
  font-family: Inter, Arial, sans-serif;
}
.logo {
  display: flex;
  align-items: center;
  gap: 18px;
}
svg { width: 100px; height: 100px; }
.brand-name {
  font-size: 30px;
  font-weight: 850;
  letter-spacing: -1.5px;
  color: black;
}
.brand-name span {
  background: linear-gradient(100deg,#00e5ff,#9361ff);
  -webkit-background-clip: text;
  background-clip: text;
  color: transparent;
}
.tagline {
  margin-top: 7px;
  color: #8996b5;
  font-size: 10px;
  font-weight: 700;
  letter-spacing: 4px;
}
</style>
</head>
<body>
<div class="logo">
  <svg viewBox="0 0 100 100" fill="none">
    <defs>
      <linearGradient id="g" x1="10" y1="10" x2="90" y2="90">
        <stop stop-color="#00e5ff"/>
        <stop offset="1" stop-color="#9361ff"/>
      </linearGradient>
    </defs>
    <path d="M50 4 91 27V73L50 96 9 73V27L50 4Z"
          fill="#111a30" stroke="url(#g)" stroke-width="2.5"/>
    <path d="m35 33-15 17 15 17M65 33l15 17-15 17"
          stroke="url(#g)" stroke-width="5"
          stroke-linecap="round" stroke-linejoin="round"/>
    <path d="m50 23 6 21 21 6-21 6-6 21-6-21-21-6 21-6 6-21Z"
          fill="url(#g)"/>
    <circle cx="50" cy="50" r="5" fill="white"/>
  </svg>
  <div>
    <div class="brand-name">LFM <span>CodeEnhanced</span></div>
    <div class="tagline">AI · CODE · CREATE</div>
  </div>
</div>
</body>
</html>


Currently has 3 quantizations 
--
| Quantization | Quality |
|----------------|--------|
| BF16 | Highest quality <span style="background-color: #12CA12; color: black; padding: 2px 8px; border-radius: 10px; font-size: 12px; font-weight: bold;">Recommended</span>
| Q4_KM | Mid Quality |
| Q8 | Also good |

---

This model was generated by deepseek V4.1 FLASH by changing model weights.

All credits to  <b>liquid-ai, deepseek & weight space programming research paper</b>

---