Severian commited on
Commit
6cdfe69
1 Parent(s): 02a1748

Upload JambaForCausalLM

Browse files
README.md CHANGED
@@ -1,10 +1,10 @@
1
  ---
2
  license: mit
 
 
3
  datasets:
4
  - teknium/OpenHermes-2.5
5
  pipeline_tag: text-generation
6
- tags:
7
- - jamba
8
  ---
9
 
10
  # PLACEHOLDER - Currently training. This is highly experimental and should be viewed as purely testing right now. Jamba has been very hard to train but I wanted to see how it did on one of the best datasets we have access to. I believe in transparent development so all *best* working iterations, even if they are a bit wonky, will be pushed here
 
1
  ---
2
  license: mit
3
+ tags:
4
+ - jamba
5
  datasets:
6
  - teknium/OpenHermes-2.5
7
  pipeline_tag: text-generation
 
 
8
  ---
9
 
10
  # PLACEHOLDER - Currently training. This is highly experimental and should be viewed as purely testing right now. Jamba has been very hard to train but I wanted to see how it did on one of the best datasets we have access to. I believe in transparent development so all *best* working iterations, even if they are a bit wonky, will be pushed here
config.json ADDED
@@ -0,0 +1,66 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "ai21labs/Jamba-v0.1",
3
+ "architectures": [
4
+ "JambaForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "attn_layer_offset": 4,
8
+ "attn_layer_period": 8,
9
+ "auto_map": {
10
+ "AutoConfig": "ai21labs/Jamba-v0.1--configuration_jamba.JambaConfig",
11
+ "AutoModel": "ai21labs/Jamba-v0.1--modeling_jamba.JambaModel",
12
+ "AutoModelForCausalLM": "ai21labs/Jamba-v0.1--modeling_jamba.JambaForCausalLM",
13
+ "AutoModelForSequenceClassification": "ai21labs/Jamba-v0.1--model.JambaForSequenceClassification"
14
+ },
15
+ "bos_token_id": 1,
16
+ "calc_logits_for_entire_prompt": false,
17
+ "eos_token_id": 2,
18
+ "expert_layer_offset": 1,
19
+ "expert_layer_period": 2,
20
+ "hidden_act": "silu",
21
+ "hidden_size": 4096,
22
+ "initializer_range": 0.02,
23
+ "intermediate_size": 14336,
24
+ "mamba_conv_bias": true,
25
+ "mamba_d_conv": 4,
26
+ "mamba_d_state": 16,
27
+ "mamba_dt_rank": 256,
28
+ "mamba_expand": 2,
29
+ "mamba_inner_layernorms": true,
30
+ "mamba_proj_bias": false,
31
+ "model_type": "jamba",
32
+ "n_ctx": 262144,
33
+ "num_attention_heads": 32,
34
+ "num_experts": 16,
35
+ "num_experts_per_tok": 2,
36
+ "num_hidden_layers": 32,
37
+ "num_key_value_heads": 8,
38
+ "output_router_logits": false,
39
+ "pad_token_id": 0,
40
+ "quantization_config": {
41
+ "_load_in_4bit": true,
42
+ "_load_in_8bit": false,
43
+ "bnb_4bit_compute_dtype": "float32",
44
+ "bnb_4bit_quant_storage": "uint8",
45
+ "bnb_4bit_quant_type": "fp4",
46
+ "bnb_4bit_use_double_quant": false,
47
+ "llm_int8_enable_fp32_cpu_offload": false,
48
+ "llm_int8_has_fp16_weight": false,
49
+ "llm_int8_skip_modules": [
50
+ "mamba"
51
+ ],
52
+ "llm_int8_threshold": 6.0,
53
+ "load_in_4bit": true,
54
+ "load_in_8bit": false,
55
+ "quant_method": "bitsandbytes"
56
+ },
57
+ "rms_norm_eps": 1e-06,
58
+ "router_aux_loss_coef": 0.001,
59
+ "sliding_window": null,
60
+ "tie_word_embeddings": false,
61
+ "torch_dtype": "bfloat16",
62
+ "transformers_version": "4.40.0.dev0",
63
+ "use_cache": true,
64
+ "use_mamba_kernels": true,
65
+ "vocab_size": 65536
66
+ }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.40.0.dev0"
7
+ }
model-00001-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc887ecb74cdcab4dd760b13741e84c495be39eb311c9ae4ca181f0d3a7b5596
3
+ size 4873003499
model-00002-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:01a96b353c80cb781570d786a1ecd0c78ab6a530b77471040b6c2a53aca594aa
3
+ size 4984623421
model-00003-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c17daf2393e7fc4e1ea4ec9b357ed33d0a70b7fc24b5376a226bbd880e46147d
3
+ size 4983413929
model-00004-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9492aa2ce2043f83e4c18075552a17b31cf33b533387bb52e15e5342660c10ea
3
+ size 4997975699
model-00005-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:867fe81979ee53424e8e9b901830c206964fae0705b8d1f2a4bc332534e2b17b
3
+ size 4983414017
model-00006-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4b11b8dc7e8642250f55cb44c427264d84eeae1c8b9d6f33426fbcbb0e267fb4
3
+ size 4975844938
model-00007-of-00007.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bf06e9efbea03f7189f806262b8174b0c27372de7b7ce912b5b2db9ba5682bf4
3
+ size 3842901948
model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff