Model comparison
TinyLlama-1.1B-Chat-v1.0 vs HyperCLOVAX-SEED-Text-Instruct-1.5B
At Q4_K_M, TinyLlama-1.1B-Chat-v1.0 is the smaller download — 667,814,880 bytes against 1,133,974,368.
From the file· summed bytes, KV per layer
Side by side
| TinyLlama-1.1B-Chat-v1.0 | HyperCLOVAX-SEED-Text-Instruct-1.5B | |
|---|---|---|
| Parameters | 1.1B | 1.6B |
| Architecture | llama | llama |
| Layers | 22 | — |
| Native context | 2,048 | — |
| Mixture of experts | no | no |
| Quantizations published | 26 | 1 |
| Smallest quantization | 0.45 GiB | 1.06 GiB |
| Q4_K_M | 0.62 GiB | 1.06 GiB |
| Licence | apache-2.0 | other |
KV cache by context
the term that decides long-context viability
| Context | TinyLlama-1.1B-Chat-v1.0 | HyperCLOVAX-SEED-Text-Instruct-1.5B | Ratio |
|---|---|---|---|
| 4,096 | 0.09 GiB | — | — |
| 8,192 | 0.17 GiB | — | — |
| 16,384 | 0.34 GiB | — | — |
| 32,768 | 0.69 GiB | — | — |
| 65,536 | 1.38 GiB | — | — |
| 131,072 | 2.75 GiB | — | — |