Re-tag as text-only causal LM (README.md)
Browse files
README.md
CHANGED
|
@@ -12,6 +12,8 @@ tags:
|
|
| 12 |
MLX quantization of [empero-ai/Qwythos-9B-Claude-Mythos-5-1M](https://huggingface.co/empero-ai/Qwythos-9B-Claude-Mythos-5-1M) for Apple Silicon.
|
| 13 |
|
| 14 |
> **Note — text tower only.** The source model is a **Qwen3.5-VL multimodal** model (`Qwen3_5ForConditionalGeneration`, with a vision encoder). This MLX conversion contains **only the text/language tower** — the vision encoder weights are **not** included, so this is a **text-only** model and does not accept image or video input. The text reasoning the original is benchmarked for (GSM8K, MMLU) is unaffected.
|
|
|
|
|
|
|
| 15 |
|
| 16 |
**Variant**: Block float MX FP8
|
| 17 |
**Disk size**: 8826 MB
|
|
|
|
| 12 |
MLX quantization of [empero-ai/Qwythos-9B-Claude-Mythos-5-1M](https://huggingface.co/empero-ai/Qwythos-9B-Claude-Mythos-5-1M) for Apple Silicon.
|
| 13 |
|
| 14 |
> **Note — text tower only.** The source model is a **Qwen3.5-VL multimodal** model (`Qwen3_5ForConditionalGeneration`, with a vision encoder). This MLX conversion contains **only the text/language tower** — the vision encoder weights are **not** included, so this is a **text-only** model and does not accept image or video input. The text reasoning the original is benchmarked for (GSM8K, MMLU) is unaffected.
|
| 15 |
+
>
|
| 16 |
+
> Because it is text-only, the config is tagged as a causal LM (`architectures: ["Qwen3_5ForCausalLM"]`, with the vision/image/video token ids removed) so it loads via the standard MLX **LLM** path (mlx-lm, LM Studio) rather than a vision-language loader.
|
| 17 |
|
| 18 |
**Variant**: Block float MX FP8
|
| 19 |
**Disk size**: 8826 MB
|