GLM-5 · Token embedding · OpenModelAtlas

Token embedding

learnedself.embed_tokens(input_ids)

Look up a learned feature vector for each token ID.

{\htmlData{light=embedded embedding}{x_0}}=E[{\htmlData{light=token-ids tokenize}{I}}]

in: $I$ [B batch, T positions]
out: $x_0$ [B batch, T positions, 6144 d]
parameters: 951,582,720

Parameters

951,582,720 parameters per instance.

Exact per-instance accounting
Component	Calculation	Parameters
Token embedding	$154{,}880 \times 6{,}144$	951,582,720
Per-instance total		951,582,720

Derived from the configured table dimensions; the model total matches the checkpoint’s tensors.

Codemodeling_glm_moe_dsa.py 669, 696–697 · config.json 1–59

modeling_glm_moe_dsa.py 669

        self.embed_tokens = nn.Embedding(config.vocab_size, config.hidden_size, self.padding_idx)

modeling_glm_moe_dsa.py 696–697

        if inputs_embeds is None:
            inputs_embeds: torch.Tensor = self.embed_tokens(input_ids)

config.json 1–59

{
  "architectures": [
    "GlmMoeDsaForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "dtype": "bfloat16",
  "eos_token_id": [
    154820,
    154827,
    154829
  ],
  "ep_size": 1,
  "first_k_dense_replace": 3,
  "hidden_act": "silu",
  "head_dim": 64,
  "hidden_size": 6144,
  "index_head_dim": 128,
  "index_n_heads": 32,
  "index_topk": 2048,
  "indexer_rope_interleave": true,
  "initializer_range": 0.02,
  "intermediate_size": 12288,
  "kv_lora_rank": 512,
  "max_position_embeddings": 202752,
  "moe_intermediate_size": 2048,
  "moe_layer_freq": 1,
  "model_type": "glm_moe_dsa",
  "n_group": 1,
  "n_routed_experts": 256,
  "n_shared_experts": 1,
  "norm_topk_prob": true,
  "num_attention_heads": 64,
  "num_experts_per_tok": 8,
  "num_hidden_layers": 78,
  "num_key_value_heads": 64,
  "num_nextn_predict_layers": 1,
  "pad_token_id": 154820,
  "pretraining_tp": 1,
  "q_lora_rank": 2048,
  "qk_head_dim": 256,
  "qk_nope_head_dim": 192,
  "qk_rope_head_dim": 64,
  "rms_norm_eps": 1e-05,
  "rope_interleave": true,
  "rope_parameters": {
    "rope_theta": 1000000,
    "rope_type": "default"
  },
  "routed_scaling_factor": 2.5,
  "scoring_func": "sigmoid",
  "tie_word_embeddings": false,
  "topk_group": 1,
  "topk_method": "noaux_tc",
  "transformers_version": "5.0.2.dev0",
  "use_cache": true,
  "v_head_dim": 256,
  "vocab_size": 154880
}

Sources2

Direct evidence

The token embedding is a learned table of vocabulary × width rows, looked up by token ID.

embedding · huggingface/transformers / modeling_glm_moe_dsa.py
Pinned revision 5e4d6304de55 ↗ · lines 669–697

Read code excerpt

        self.embed_tokens = nn.Embedding(config.vocab_size, config.hidden_size, self.padding_idx)

… lines 670–695 omitted …

        if inputs_embeds is None:
            inputs_embeds: torch.Tensor = self.embed_tokens(input_ids)

Direct evidence

The released configuration: width, 78 layers of which 3 dense, 64 heads, 256 routed experts and one shared, routing (sigmoid scores, one expert group, top-8, route scale 2.5), MLA ranks and head widths, the indexer (32 heads of 128, top 2,048, interleaved RoPE), vocabulary, default RoPE with θ = 1,000,000, and one multi-token-prediction layer.

config · zai-org/GLM-5 / config.json
Pinned revision c183ef8c61fa ↗ · lines 1–59

Read code excerpt

{
  "architectures": [
    "GlmMoeDsaForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "dtype": "bfloat16",
  "eos_token_id": [
    154820,
    154827,
    154829
  ],
  "ep_size": 1,
  "first_k_dense_replace": 3,
  "hidden_act": "silu",
  "head_dim": 64,
  "hidden_size": 6144,
  "index_head_dim": 128,
  "index_n_heads": 32,
  "index_topk": 2048,
  "indexer_rope_interleave": true,
  "initializer_range": 0.02,
  "intermediate_size": 12288,
  "kv_lora_rank": 512,
  "max_position_embeddings": 202752,
  "moe_intermediate_size": 2048,
  "moe_layer_freq": 1,
  "model_type": "glm_moe_dsa",
  "n_group": 1,
  "n_routed_experts": 256,
  "n_shared_experts": 1,
  "norm_topk_prob": true,
  "num_attention_heads": 64,
  "num_experts_per_tok": 8,
  "num_hidden_layers": 78,
  "num_key_value_heads": 64,
  "num_nextn_predict_layers": 1,
  "pad_token_id": 154820,
  "pretraining_tp": 1,
  "q_lora_rank": 2048,
  "qk_head_dim": 256,
  "qk_nope_head_dim": 192,
  "qk_rope_head_dim": 64,
  "rms_norm_eps": 1e-05,
  "rope_interleave": true,
  "rope_parameters": {
    "rope_theta": 1000000,
    "rope_type": "default"
  },
  "routed_scaling_factor": 2.5,
  "scoring_func": "sigmoid",
  "tie_word_embeddings": false,
  "topk_group": 1,
  "topk_method": "noaux_tc",
  "transformers_version": "5.0.2.dev0",
  "use_cache": true,
  "v_head_dim": 256,
  "vocab_size": 154880
}