Qwen · 2024-09 · Dense
Qwen2.5 72B
Qwen2.5 72B is a grouped-query attention (GQA) transformer released by Qwen in 2024-09, with 80 layers, hidden size 8192 and a context window of 32,768 tokens.
Layer stack
Attention ×80 · Layers 80
Key facts
| Family | Qwen |
|---|---|
| Released | 2024-09 |
| Params | 71.4 B |
| Context | 32,768 tokens |
| Attention | GQA (64:8) |
| Layers | 80 |
| Hidden | 8,192 |
| Heads | 64 |
| Vocab | 152,064 |
| Position | RoPE |
| Norm | RMSNorm |
| Activation | silu |
| Dtype | bf16 |
| Architecture class | Qwen2ForCausalLM |
Architecture overview
Attention
grouped-query attention (GQA) — 64 q-heads / 8 kv-heads. RoPE θ=1,000,000.
Feed-forward / MoE
It is a dense model: all 71.4B parameters participate in computing every token.
Field-level comparison
Compared against the previous model of the same lab; where none exists, against the structurally closest model. The ratio column is this model divided by the comparison model.
Full-field comparison vs predecessor Qwen2 72B
| Model | Qwen2.5 72B | Qwen2 72B | Ratio |
|---|---|---|---|
| model_type | qwen2 | qwen2 | ＝ |
| architectures | Qwen2ForCausalLM | Qwen2ForCausalLM | ＝ |
| hidden_size | 8192 | 8192 | ≈1 |
| num_hidden_layers | 80 | 80 | ≈1 |
| num_attention_heads | 64 | 64 | ≈1 |
| num_key_value_heads | 8 | 8 | ≈1 |
| intermediate_size | 29568 | 29568 | ≈1 |
| hidden_act | silu | silu | ＝ |
| rope_theta | 1000000 | 1000000 | ≈1 |
| max_position_embeddings | 32768 | 32768 | ≈1 |
| sliding_window | 131072 | 131072 | ≈1 |
| rms_norm_eps | 0.000001 | 0.000001 | ≈1 |
| vocab_size | 152064 | 152064 | ≈1 |
| torch_dtype | bfloat16 | bfloat16 | ＝ |
| attention_dropout | 0 | 0 | ＝ |
| bos_token_id | 151643 | 151643 | ≈1 |
| eos_token_id | 151645 | 151645 | ≈1 |
| initializer_range | 0.02 | 0.02 | ≈1 |
| max_window_layers | 70 | 80 | ×0.875 |
| tie_word_embeddings | false | false | ＝ |
| transformers_version | 4.43.1 | 4.40.1 | ≠ |
| use_cache | true | true | ＝ |
| use_sliding_window | false | false | ＝ |
Most similar architectures
- Qwen2 72B (Qwen, 2024-06) 1.000
- Llama 3.1 70B (Meta, 2024-07) 0.999
- Qwen3 32B (Qwen, 2025-04) 0.995
- Qwen3.6 27B (Qwen, 2026-04) 0.992
- Qwen3.8 27B (Qwen, 2026-08) 0.992
- Qwen3 8B (Qwen, 2025-04) 0.977
Similarity values range from 0 (no shared categorical features) to 1 (identical profiles).
Raw config fields
23 fields
| architectures | Qwen2ForCausalLM |
|---|---|
| attention_dropout | 0 |
| bos_token_id | 151643 |
| eos_token_id | 151645 |
| hidden_act | silu |
| hidden_size | 8192 |
| initializer_range | 0.02 |
| intermediate_size | 29568 |
| max_position_embeddings | 32768 |
| max_window_layers | 70 |
| model_type | qwen2 |
| num_attention_heads | 64 |
| num_hidden_layers | 80 |
| num_key_value_heads | 8 |
| rms_norm_eps | 0.000001 |
| rope_theta | 1000000 |
| sliding_window | 131072 |
| tie_word_embeddings | false |
| torch_dtype | bfloat16 |
| transformers_version | 4.43.1 |
| use_cache | true |
| use_sliding_window | false |
| vocab_size | 152064 |