convert : fix FP8 qwen4exp checkpoints (n-gram table scale, block-scaled FP8) (#32476)

* convert : fix FP8 n-gram table export for qwen4exp

* convert : dequantize block-scaled FP8 weight_scale_inv in ModelOpt checkpoints
This commit is contained in:
Aman Gupta authored and GitHub committed 2026-11-17 22:25:01 +08:00
1 parent 7f106b5459
commit d3064fddce
6 files changed
+160 -8

No files matched your search

+1
View File
@@ -571,6 +571,7 @@ void llama_model_saver::add_tensors_from_model() {
add_tensor(model->hc_head_base);
add_tensor(model->hc_head_scale);
add_tensor(model->per_layer_tok_embd);
add_tensor(model->per_layer_tok_embd_s);
add_tensor(model->per_layer_model_proj);
add_tensor(model->per_layer_proj_norm);
add_tensor(model->value_embd);