Models
GitHub
Discord
Docs
Cloud
Sign in
Download
Models
Download
GitHub
Discord
Docs
Cloud
Sign in
cnmoro
/
LFM2-2.6B
:q4_k_m
82
Downloads
Updated
3 weeks ago
Cancel
tools
LFM2-2.6B:q4_k_m
...
/
model
384bc877b6c3 · 1.6GB
Metadata
general.architecture
lfm2
lfm2
general.file_type
Q4_K_M
Q4_K_M
lfm2.attention.head_count
32
32
lfm2.attention.head_count_kv
[0, 0, 8, 0, 0, ...]
[0, 0, 8, 0, 0, ...]
lfm2.attention.layer_norm_rms_epsilon
1e-05
1e-05
lfm2.block_count
30
30
lfm2.context_length
128000
128000
lfm2.embedding_length
2048
2048
lfm2.feed_forward_length
10752
10752
lfm2.rope.freq_base
1e+06
1e+06
lfm2.shortconv.l_cache
3
3
lfm2.vocab_size
65536
65536
tokenizer.ggml.add_bos_token
true
true
tokenizer.ggml.add_eos_token
false
false
tokenizer.ggml.add_sep_token
false
false
tokenizer.ggml.bos_token_id
1
1
tokenizer.ggml.eos_token_id
7
7
tokenizer.ggml.merges
[Ċ Ċ, Ċ ĊĊ, ĊĊ Ċ, Ċ ĊĊĊ, ĊĊ ĊĊ, ...]
[Ċ Ċ, Ċ ĊĊ, ĊĊ Ċ, Ċ ĊĊĊ, ĊĊ ĊĊ, ...]
tokenizer.ggml.model
gpt2
gpt2
tokenizer.ggml.padding_token_id
0
0
tokenizer.ggml.pre
lfm2
lfm2
tokenizer.ggml.token_type
[3, 3, 3, 3, 3, ...]
[3, 3, 3, 3, 3, ...]
tokenizer.ggml.tokens
[<|pad|>, <|startoftext|>, <|endoftext|>, <|fim_pre|>, <|fim_mid|>, ...]
[<|pad|>, <|startoftext|>, <|endoftext|>, <|fim_pre|>, <|fim_mid|>, ...]
Tensor
Name
Type
Shape
token_embd.weight
Q6_K
Q6_K
[2048, 65536]
blk.0
blk.0.attn_norm.weight
F32
F32
[2048]
blk.0.ffn_down.weight
Q6_K
Q6_K
[10752, 2048]
blk.0.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.0.ffn_norm.weight
F32
F32
[2048]
blk.0.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.0.shortconv.conv.weight
F32
F32
[3, 2048]
blk.0.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.0.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.1
blk.1.attn_norm.weight
F32
F32
[2048]
blk.1.ffn_down.weight
Q6_K
Q6_K
[10752, 2048]
blk.1.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.1.ffn_norm.weight
F32
F32
[2048]
blk.1.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.1.shortconv.conv.weight
F32
F32
[3, 2048]
blk.1.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.1.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.2
blk.2.attn_k.weight
Q4_K
Q4_K
[2048, 512]
blk.2.attn_k_norm.weight
F32
F32
[64]
blk.2.attn_norm.weight
F32
F32
[2048]
blk.2.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.2.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.2.attn_q_norm.weight
F32
F32
[64]
blk.2.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.2.ffn_down.weight
Q6_K
Q6_K
[10752, 2048]
blk.2.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.2.ffn_norm.weight
F32
F32
[2048]
blk.2.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.3
blk.3.attn_norm.weight
F32
F32
[2048]
blk.3.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.3.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.3.ffn_norm.weight
F32
F32
[2048]
blk.3.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.3.shortconv.conv.weight
F32
F32
[3, 2048]
blk.3.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.3.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.4
blk.4.attn_norm.weight
F32
F32
[2048]
blk.4.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.4.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.4.ffn_norm.weight
F32
F32
[2048]
blk.4.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.4.shortconv.conv.weight
F32
F32
[3, 2048]
blk.4.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.4.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.5
blk.5.attn_k.weight
Q4_K
Q4_K
[2048, 512]
blk.5.attn_k_norm.weight
F32
F32
[64]
blk.5.attn_norm.weight
F32
F32
[2048]
blk.5.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.5.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.5.attn_q_norm.weight
F32
F32
[64]
blk.5.attn_v.weight
Q4_K
Q4_K
[2048, 512]
blk.5.ffn_down.weight
Q6_K
Q6_K
[10752, 2048]
blk.5.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.5.ffn_norm.weight
F32
F32
[2048]
blk.5.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.6
blk.6.attn_norm.weight
F32
F32
[2048]
blk.6.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.6.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.6.ffn_norm.weight
F32
F32
[2048]
blk.6.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.6.shortconv.conv.weight
F32
F32
[3, 2048]
blk.6.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.6.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.7
blk.7.attn_norm.weight
F32
F32
[2048]
blk.7.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.7.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.7.ffn_norm.weight
F32
F32
[2048]
blk.7.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.7.shortconv.conv.weight
F32
F32
[3, 2048]
blk.7.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.7.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.8
blk.8.attn_norm.weight
F32
F32
[2048]
blk.8.ffn_down.weight
Q6_K
Q6_K
[10752, 2048]
blk.8.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.8.ffn_norm.weight
F32
F32
[2048]
blk.8.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.8.shortconv.conv.weight
F32
F32
[3, 2048]
blk.8.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.8.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.9
blk.9.attn_k.weight
Q4_K
Q4_K
[2048, 512]
blk.9.attn_k_norm.weight
F32
F32
[64]
blk.9.attn_norm.weight
F32
F32
[2048]
blk.9.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.9.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.9.attn_q_norm.weight
F32
F32
[64]
blk.9.attn_v.weight
Q4_K
Q4_K
[2048, 512]
blk.9.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.9.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.9.ffn_norm.weight
F32
F32
[2048]
blk.9.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.10
blk.10.attn_norm.weight
F32
F32
[2048]
blk.10.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.10.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.10.ffn_norm.weight
F32
F32
[2048]
blk.10.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.10.shortconv.conv.weight
F32
F32
[3, 2048]
blk.10.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.10.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.11
blk.11.attn_norm.weight
F32
F32
[2048]
blk.11.ffn_down.weight
Q6_K
Q6_K
[10752, 2048]
blk.11.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.11.ffn_norm.weight
F32
F32
[2048]
blk.11.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.11.shortconv.conv.weight
F32
F32
[3, 2048]
blk.11.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.11.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.12
blk.12.attn_norm.weight
F32
F32
[2048]
blk.12.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.12.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.12.ffn_norm.weight
F32
F32
[2048]
blk.12.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.12.shortconv.conv.weight
F32
F32
[3, 2048]
blk.12.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.12.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.13
blk.13.attn_k.weight
Q4_K
Q4_K
[2048, 512]
blk.13.attn_k_norm.weight
F32
F32
[64]
blk.13.attn_norm.weight
F32
F32
[2048]
blk.13.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.13.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.13.attn_q_norm.weight
F32
F32
[64]
blk.13.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.13.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.13.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.13.ffn_norm.weight
F32
F32
[2048]
blk.13.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.14
blk.14.attn_norm.weight
F32
F32
[2048]
blk.14.ffn_down.weight
Q6_K
Q6_K
[10752, 2048]
blk.14.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.14.ffn_norm.weight
F32
F32
[2048]
blk.14.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.14.shortconv.conv.weight
F32
F32
[3, 2048]
blk.14.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.14.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.15
blk.15.attn_norm.weight
F32
F32
[2048]
blk.15.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.15.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.15.ffn_norm.weight
F32
F32
[2048]
blk.15.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.15.shortconv.conv.weight
F32
F32
[3, 2048]
blk.15.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.15.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.16
blk.16.attn_norm.weight
F32
F32
[2048]
blk.16.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.16.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.16.ffn_norm.weight
F32
F32
[2048]
blk.16.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.16.shortconv.conv.weight
F32
F32
[3, 2048]
blk.16.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.16.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.17
blk.17.attn_k.weight
Q4_K
Q4_K
[2048, 512]
blk.17.attn_k_norm.weight
F32
F32
[64]
blk.17.attn_norm.weight
F32
F32
[2048]
blk.17.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.17.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.17.attn_q_norm.weight
F32
F32
[64]
blk.17.attn_v.weight
Q4_K
Q4_K
[2048, 512]
blk.17.ffn_down.weight
Q6_K
Q6_K
[10752, 2048]
blk.17.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.17.ffn_norm.weight
F32
F32
[2048]
blk.17.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.18
blk.18.attn_norm.weight
F32
F32
[2048]
blk.18.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.18.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.18.ffn_norm.weight
F32
F32
[2048]
blk.18.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.18.shortconv.conv.weight
F32
F32
[3, 2048]
blk.18.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.18.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.19
blk.19.attn_norm.weight
F32
F32
[2048]
blk.19.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.19.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.19.ffn_norm.weight
F32
F32
[2048]
blk.19.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.19.shortconv.conv.weight
F32
F32
[3, 2048]
blk.19.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.19.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.20
blk.20.attn_norm.weight
F32
F32
[2048]
blk.20.ffn_down.weight
Q6_K
Q6_K
[10752, 2048]
blk.20.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.20.ffn_norm.weight
F32
F32
[2048]
blk.20.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.20.shortconv.conv.weight
F32
F32
[3, 2048]
blk.20.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.20.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.21
blk.21.attn_k.weight
Q4_K
Q4_K
[2048, 512]
blk.21.attn_k_norm.weight
F32
F32
[64]
blk.21.attn_norm.weight
F32
F32
[2048]
blk.21.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.21.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.21.attn_q_norm.weight
F32
F32
[64]
blk.21.attn_v.weight
Q4_K
Q4_K
[2048, 512]
blk.21.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.21.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.21.ffn_norm.weight
F32
F32
[2048]
blk.21.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.22
blk.22.attn_norm.weight
F32
F32
[2048]
blk.22.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.22.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.22.ffn_norm.weight
F32
F32
[2048]
blk.22.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.22.shortconv.conv.weight
F32
F32
[3, 2048]
blk.22.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.22.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.23
blk.23.attn_norm.weight
F32
F32
[2048]
blk.23.ffn_down.weight
Q6_K
Q6_K
[10752, 2048]
blk.23.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.23.ffn_norm.weight
F32
F32
[2048]
blk.23.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.23.shortconv.conv.weight
F32
F32
[3, 2048]
blk.23.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.23.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.24
blk.24.attn_k.weight
Q4_K
Q4_K
[2048, 512]
blk.24.attn_k_norm.weight
F32
F32
[64]
blk.24.attn_norm.weight
F32
F32
[2048]
blk.24.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.24.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.24.attn_q_norm.weight
F32
F32
[64]
blk.24.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.24.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.24.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.24.ffn_norm.weight
F32
F32
[2048]
blk.24.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.25
blk.25.attn_norm.weight
F32
F32
[2048]
blk.25.ffn_down.weight
Q4_K
Q4_K
[10752, 2048]
blk.25.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.25.ffn_norm.weight
F32
F32
[2048]
blk.25.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.25.shortconv.conv.weight
F32
F32
[3, 2048]
blk.25.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.25.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.26
blk.26.attn_norm.weight
F32
F32
[2048]
blk.26.ffn_down.weight
Q6_K
Q6_K
[10752, 2048]
blk.26.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.26.ffn_norm.weight
F32
F32
[2048]
blk.26.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.26.shortconv.conv.weight
F32
F32
[3, 2048]
blk.26.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.26.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.27
blk.27.attn_k.weight
Q4_K
Q4_K
[2048, 512]
blk.27.attn_k_norm.weight
F32
F32
[64]
blk.27.attn_norm.weight
F32
F32
[2048]
blk.27.attn_output.weight
Q4_K
Q4_K
[2048, 2048]
blk.27.attn_q.weight
Q4_K
Q4_K
[2048, 2048]
blk.27.attn_q_norm.weight
F32
F32
[64]
blk.27.attn_v.weight
Q6_K
Q6_K
[2048, 512]
blk.27.ffn_down.weight
Q6_K
Q6_K
[10752, 2048]
blk.27.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.27.ffn_norm.weight
F32
F32
[2048]
blk.27.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.28
blk.28.attn_norm.weight
F32
F32
[2048]
blk.28.ffn_down.weight
Q6_K
Q6_K
[10752, 2048]
blk.28.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.28.ffn_norm.weight
F32
F32
[2048]
blk.28.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.28.shortconv.conv.weight
F32
F32
[3, 2048]
blk.28.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.28.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
blk.29
blk.29.attn_norm.weight
F32
F32
[2048]
blk.29.ffn_down.weight
Q6_K
Q6_K
[10752, 2048]
blk.29.ffn_gate.weight
Q4_K
Q4_K
[2048, 10752]
blk.29.ffn_norm.weight
F32
F32
[2048]
blk.29.ffn_up.weight
Q4_K
Q4_K
[2048, 10752]
blk.29.shortconv.conv.weight
F32
F32
[3, 2048]
blk.29.shortconv.in_proj.weight
Q4_K
Q4_K
[2048, 6144]
blk.29.shortconv.out_proj.weight
Q4_K
Q4_K
[2048, 2048]
token_embd_norm.weight
F32
F32
[2048]