Update README.md
Browse files
README.md
CHANGED
|
@@ -49,11 +49,11 @@ In this benchmark table we can see Fijik2.0 performing competitively with simila
|
|
| 49 |
```
|
| 50 |
Granite 4.0 350M base
|
| 51 |
\ /
|
| 52 |
-
Continual pre-training1 (~
|
| 53 |
\ /
|
| 54 |
Continual pre-training2 (~1B tokens, aug 2025 knowledge cutoff, higher context)
|
| 55 |
\ /
|
| 56 |
-
Supervised fine-tuning (~
|
| 57 |
\ /
|
| 58 |
Final uploaded model (best checkpoint chosen)
|
| 59 |
```
|
|
|
|
| 49 |
```
|
| 50 |
Granite 4.0 350M base
|
| 51 |
\ /
|
| 52 |
+
Continual pre-training1 (~2B tokens, august 2025 knowledge cutoff, low context)
|
| 53 |
\ /
|
| 54 |
Continual pre-training2 (~1B tokens, aug 2025 knowledge cutoff, higher context)
|
| 55 |
\ /
|
| 56 |
+
Supervised fine-tuning (~2B tokens, data from feb 2026, mixed chat, agentic, web search, code, tool use, reasoning)
|
| 57 |
\ /
|
| 58 |
Final uploaded model (best checkpoint chosen)
|
| 59 |
```
|