anuj0456 commited on
Commit
c0b74e4
·
verified ·
1 Parent(s): 7c070a4

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +0 -12
README.md CHANGED
@@ -19,8 +19,6 @@ library_name: transformers
19
 
20
  This is a **base scientific language model** (not instruction-tuned).
21
 
22
- ---
23
-
24
  ## Overview
25
 
26
  KiteFish-A1-1.5B explores what it takes to train a domain-specialized scientific language model directly from structured LaTeX archives.
@@ -35,8 +33,6 @@ KiteFish-A1-1.5B explores what it takes to train a domain-specialized scientific
35
 
36
  The focus of this project is *scientific language modeling robustness*, not benchmark optimization.
37
 
38
- ---
39
-
40
  ## Model Architecture
41
 
42
  - 24 Transformer layers
@@ -57,8 +53,6 @@ The focus of this project is *scientific language modeling robustness*, not benc
57
 
58
  **Validation Perplexity:** ~4.2 (held-out scientific corpus)
59
 
60
- ---
61
-
62
  ## Intended Use
63
 
64
  KiteFish-A1-1.5B is suitable for:
@@ -76,8 +70,6 @@ It is **not optimized for:**
76
  - General conversational AI
77
  - Benchmark leaderboard performance
78
 
79
- ---
80
-
81
  ## Performance Notes
82
 
83
  This model was trained under moderate compute constraints and without instruction tuning or alignment stages.
@@ -92,8 +84,6 @@ Observed characteristics:
92
 
93
  Performance improves significantly with supervised fine-tuning (SFT), LoRA adaptation, or domain-specific instruction tuning.
94
 
95
- ---
96
-
97
  ## Limitations
98
 
99
  - Not instruction-tuned
@@ -105,8 +95,6 @@ Performance improves significantly with supervised fine-tuning (SFT), LoRA adapt
105
 
106
  This release is intended primarily for research and experimentation.
107
 
108
- ---
109
-
110
  ## Example Usage
111
 
112
  ```python
 
19
 
20
  This is a **base scientific language model** (not instruction-tuned).
21
 
 
 
22
  ## Overview
23
 
24
  KiteFish-A1-1.5B explores what it takes to train a domain-specialized scientific language model directly from structured LaTeX archives.
 
33
 
34
  The focus of this project is *scientific language modeling robustness*, not benchmark optimization.
35
 
 
 
36
  ## Model Architecture
37
 
38
  - 24 Transformer layers
 
53
 
54
  **Validation Perplexity:** ~4.2 (held-out scientific corpus)
55
 
 
 
56
  ## Intended Use
57
 
58
  KiteFish-A1-1.5B is suitable for:
 
70
  - General conversational AI
71
  - Benchmark leaderboard performance
72
 
 
 
73
  ## Performance Notes
74
 
75
  This model was trained under moderate compute constraints and without instruction tuning or alignment stages.
 
84
 
85
  Performance improves significantly with supervised fine-tuning (SFT), LoRA adaptation, or domain-specific instruction tuning.
86
 
 
 
87
  ## Limitations
88
 
89
  - Not instruction-tuned
 
95
 
96
  This release is intended primarily for research and experimentation.
97
 
 
 
98
  ## Example Usage
99
 
100
  ```python