Update README.md
Browse files
README.md
CHANGED
|
@@ -17,6 +17,9 @@ base_model:
|
|
| 17 |
pipeline_tag: text-generation
|
| 18 |
---
|
| 19 |
|
|
|
|
|
|
|
|
|
|
| 20 |
# Model Description
|
| 21 |
|
| 22 |
**KafkaLM‑15B‑Base** is a 15‑billion‑parameter, sparsity‑aware language model distilled from *Mistral‑Small‑24B‑Base‑2501* and further post trained (SFT + DPO + GRPO /w verifiable rewards).
|
|
@@ -86,7 +89,8 @@ Up to 40 % parameter reduction (24 B → 15 B) delivers 2× lower TTFT
|
|
| 86 |
| Tokens / s | 579 | **812** | +40% |
|
| 87 |
|
| 88 |
|
| 89 |
-
|
|
|
|
| 90 |
|
| 91 |
### Training scalability (distillation run, MI300A cluster)
|
| 92 |
|
|
|
|
| 17 |
pipeline_tag: text-generation
|
| 18 |
---
|
| 19 |
|
| 20 |
+
<img src="https://cdn-uploads.huggingface.co/production/uploads/645ded34a45b4182d7f5c385/EgsjPDWd37LjAtamiICxk.png" width="480" height="480" alt="image/png">
|
| 21 |
+
|
| 22 |
+
|
| 23 |
# Model Description
|
| 24 |
|
| 25 |
**KafkaLM‑15B‑Base** is a 15‑billion‑parameter, sparsity‑aware language model distilled from *Mistral‑Small‑24B‑Base‑2501* and further post trained (SFT + DPO + GRPO /w verifiable rewards).
|
|
|
|
| 89 |
| Tokens / s | 579 | **812** | +40% |
|
| 90 |
|
| 91 |
|
| 92 |
+
<img src="https://cdn-uploads.huggingface.co/production/uploads/645ded34a45b4182d7f5c385/4rDhaeC-1GMj6KWbB27f9.png" width="480" height="480" alt="image/png">
|
| 93 |
+
|
| 94 |
|
| 95 |
### Training scalability (distillation run, MI300A cluster)
|
| 96 |
|