Update README.md and config files

.gitattributes

@@ -33,6 +33,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
-assets/K-EXAONE_Symbol_3d.png filter=lfs diff=lfs merge=lfs -text
+assets/K-EXAONE_logo_gray.png filter=lfs diff=lfs merge=lfs -text
 assets/main_figure.png filter=lfs diff=lfs merge=lfs -text
 tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -20,7 +20,7 @@ library_name: transformers
 <br>
 <br>
 <p align="center">
-<img src="assets/K-EXAONE_Symbol_3d.png" width="400">
+<img src="assets/K-EXAONE_logo_gray.png" width="400">
 <br>
 <br>
 <br>
@@ -51,6 +51,8 @@ library_name: transformers
 
 <br>
 
+# K-EXAONE-236B-A23B
+
 ## Introduction
 
 We introduce **K-EXAONE**, a large-scale multilingual language model developed by LG AI Research. Built using a Mixture-of-Experts architecture, K-EXAONE features **236 billion total** parameters, with **23 billion active** during inference. Performance evaluations across various benchmarks demonstrate that K-EXAONE excels in reasoning, agentic capabilities, general knowledge, multilingual understanding, and long-context processing.
@@ -382,22 +384,22 @@ Until the libraries officially support K-EXAONE, you need to install the require
 
 #### Transformers
 
-You can install the latest version of Transformers with support for EXAONE-MoE architecture from [this repository](https://github.com/Aim-Highest/transformers).
+You can install the latest version of Transformers with support for EXAONE-MoE architecture from [this repository](https://github.com/nuxlear/transformers/tree/add-exaone-moe).
 The base version of Transformers is `5.0.0rc1`, so it might be helpful to check [the migration guide](https://github.com/huggingface/transformers/blob/main/MIGRATION_GUIDE_V5.md) from the Transformers library.
 
 #### vLLM
 
 You should install both Transformers and vLLM to use K-EXAONE model on vLLM server.
-You can install the latest version of vLLM with support for EXAONE-MoE architecture from [this repository](https://github.com/Aim-Highest/vllm/tree/add-exaone-moe). 
+You can install the latest version of vLLM with support for EXAONE-MoE architecture from [this repository](https://github.com/lkm2835/vllm/tree/add-exaone-moe). 
 
 #### SGLang
 
 You should install both Transformers and SGLang to use K-EXAONE model on SGLang server.
-You can install the latest version of SGLang with support for EXAONE-MoE architecture from [this repository](https://github.com/Aim-Highest/sglang).
+You can install the latest version of SGLang with support for EXAONE-MoE architecture from [this repository](https://github.com/xvyaward/sglang/tree/exaone_moe_official).
 
 #### llama.cpp
 
-You can install the latest version of llama.cpp with support for EXAONE-MoE architecture from [this repository](https://github.com/Aim-Highest/llama.cpp).
+You can install the latest version of llama.cpp with support for EXAONE-MoE architecture from [this repository](https://github.com/nuxlear/llama.cpp/tree/add-exaone-moe).
 Please refer to the [official build guide](https://github.com/ggml-org/llama.cpp/blob/master/docs/build.md) for details. 
 
 
@@ -438,6 +440,7 @@ generated_ids = model.generate(
     max_new_tokens=16384,
     temperature=1.0,
     top_p=0.95,
+    do_sample=True,
 )
 output_ids = generated_ids[0][input_ids['input_ids'].shape[-1]:]
 print(tokenizer.decode(output_ids, skip_special_tokens=True))
@@ -465,6 +468,7 @@ generated_ids = model.generate(
     max_new_tokens=1024,
     temperature=1.0,
     top_p=0.95,
+    do_sample=True,
 )
 output_ids = generated_ids[0][input_ids['input_ids'].shape[-1]:]
 print(tokenizer.decode(output_ids, skip_special_tokens=True))
@@ -510,6 +514,7 @@ generated_ids = model.generate(
     max_new_tokens=16384,
     temperature=1.0,
     top_p=0.95,
+    do_sample=True,
 )
 output_ids = generated_ids[0][input_ids['input_ids'].shape[-1]:]
 print(tokenizer.decode(output_ids, skip_special_tokens=True))

config.json

@@ -6,7 +6,7 @@
   "bos_token_id": 1,
   "dtype": "bfloat16",
   "eos_token_id": 53,
-  "first_last_k_dense_replace": 1,
+  "first_k_dense_replace": 1,
   "head_dim": 128,
   "hidden_act": "silu",
   "hidden_size": 6144,
@@ -130,13 +130,11 @@
     "rope_type": "default"
   },
   "routed_scaling_factor": 2.5,
-  "scoring_func": "sigmoid",
   "sliding_window": 128,
   "sliding_window_pattern": "LLLG",
   "tie_word_embeddings": false,
   "tokenizer_class": "GPT2Tokenizer",
   "topk_group": 1,
-  "topk_method": "noaux_tc",
   "transformers_version": "5.0.0.dev0",
   "use_cache": true,
   "vocab_size": 153600

generation_config.json

@@ -1,6 +1,7 @@
 {
   "_from_model_config": true,
   "bos_token_id": 1,
+  "do_sample": true,
   "eos_token_id": 53,
   "pad_token_id": 0,
   "presence_penalty": 0.0,